Projects - User contributions [en]

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:59:21Z

A1675729: /* 4.Path Planning */

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

[[File:IPprocedure.png]]

Example outputs of the classification and feature
extraction phases are shown below.

[[File:IPresult.png]]

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.

A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:

𝑓(𝑛) = 𝑔(𝑛) + ℎ(𝑛)

The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

[[File:PathPlanning.png]]

== 5.Motion Control ==

5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:58:38Z

A1675729: /* 3.Image Processing */

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

[[File:IPprocedure.png]]

Example outputs of the classification and feature
extraction phases are shown below.

[[File:IPresult.png]]

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.

A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:

𝑓(𝑛) = 𝑔(𝑛) + ℎ(𝑛)

The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

== 5.Motion Control ==

5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

File:PathPlanning.png

2017-06-05T06:57:33Z

A1675729:

File:IPresult.png

2017-06-05T06:57:12Z

A1675729:

File:IPprocedure.png

2017-06-05T06:56:46Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:53:11Z

A1675729: /* 4.Path Planning */

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

Example outputs of the classification and feature
extraction phases are shown below.

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.

A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:

𝑓(𝑛) = 𝑔(𝑛) + ℎ(𝑛)

The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

== 5.Motion Control ==

5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:52:39Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

Example outputs of the classification and feature
extraction phases are shown below.

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.

A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:

𝑓(𝑛) = 𝑔(𝑛) + ℎ(𝑛)

The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

== 5.Motion Control ==

5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:51:27Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

Example outputs of the classification and feature
extraction phases are shown below.

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.

A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:

𝑓 𝑛 = 𝑔 𝑛 + ℎ(𝑛)

The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:49:21Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera

• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.

• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path

• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

== 3.Image Processing ==

'''3.1 Methods'''

1. Image Capture

2. Down-sample

3. Projective Transformation

4. Thresholding and Segmentation

5. Feature Extraction

6. Classification of Objects

7. Describe Environment for path planning

8. Provide positional feedback for control system
compensation

'''3.2 Results'''

An example output of the projective transformation is
shown below. This stage allows the image to represent a
two-dimension Cartesian plane which aids in geometric
calculation accuracy

Example outputs of the classification and feature
extraction phases are shown below.

Left: An example of calculated motion parameters. Blue
denotes the current path segment, Red indicates distance
to destination and Green indicates the current robot
bearing.

Right: The robot’s environment with only obstacles
highlighted, after the robot is automatically removed from
the frame.

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.
A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

A-star algorithm:
𝑓 𝑛 = 𝑔 𝑛 + ℎ(𝑛)
The optimal path search algorithm is based on
calculation of historic cost g(n) and heuristic cost h(n).
The optimal path nodes are found by minimizing the
total cost function f(n).

'''4.2 Results'''

Example of path calculated in the GUI using
virtually added obstacles:

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

The outcome of the final system demonstrates that the aims of the project are successfully reached. The image processing system is capable of identifying obstacles and robot and is able to monitoring the robot navigation while sending correction information to the robot. The path planning system based on A-star algorithm calculates and delivers optimal path efficiently with the given map information. The Quanser robot control system uses Simulink model to control the robot to navigate following calculated path to a satisfied level. Overall, the three main parts of the whole system cooperate well to perform a collision-free robot navigation in an environment with static obstacles.

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:42:23Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

• Microsoft Kinect V2 Camera
• Quanser Qbot2 robot

'''2.2 Software'''

• The Image Processing software locates objects and
provides robot location and bearing.
• The Path Planning software and GUI adds virtual
obstacles and calculates the optimal path
• The Motion Control Simulink Model calculates control
parameters based on calculated path, robot location
and orientation.

2.2.1 Image processing software

2.2.2 Path planning software

2.2.3 Motion control software

== 3.Image Processing ==

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.
A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

'''4.2 Result'''

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:39:39Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:SystemStructure.png]]

'''2.1Hardware'''

2.1.1 Camera

2.1.2 Quanser

'''2.2 Software'''

2.2.1 Image processing software

2.2.2 Path planning software

2.2.3 Motion control software

== 3.Image Processing ==

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.
A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

'''4.2 Result'''

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

File:SystemStructure.png

2017-06-05T06:37:20Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:34:10Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation and background'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.3 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:]]

'''2.1 Hardware'''

2.1.1 Camera

2.1.2 Quanser

'''2.2 Software'''

2.2.1 Image processing software

2.2.2 Path planning software

2.2.3 Motion control software

== 3.Image Processing ==

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.
A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

'''4.2 Result'''

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-05T06:32:28Z

A1675729:

'''General Information'''

Description: The purpose of this project is to design a robotic system for path planning in congestion area. Potential applications are in autopilot urban transportations. The robot will be located using vision measurement through a camera. The robot models are established as two-wheel discs. Virtual obstacle clusters are manually inserted into the image frame and the computer needs to calculate a suitable path to a user-defined destination. As the obstacles are considered to be physical, collision avoidance is required. The robot can be equipped with sensors and the computer captures images from the vision stream. The communication between the computer and the robots will be through WiFi. Matlab, c++, OpenGL, and OpenCV are some of the softwares, languages and libraries to choose from.

Group number:220

Students: Michael Vieceli; Yuanyang Wang; Di Zhang

Superviosrs: Professor Peng Shi & Professor Cheng-Chew Lim;

Technical advisors: Dr. Hongjun Yu & Dr. Bing Yan

== 1.Introduction ==

'''1.1 Motivation'''

Image processing techniques are widely used in industrial
automation for quality control, condition monitoring and
increasingly for motion control.
Image processing provides absolute state feedback which can
be combined with conventional position sensors to reduce the
impact of accumulative error through the course of repeated
movement. Although uncertainty introduced into the imaging
system through noise, perspective compensation and
increased computational delay, these uncertainties tend to be
constant over time.
Overhead cameras also provide the benefit of analysing an
environment for obstacles, personnel and other locomotive
bodies which would interfere with the unobstructed movement
of a robot or vehicle.
The goal of this project is to identify obstacles and a robot in a
given area, calculate a safe path and then traverse the path.
Positional feedback will also be used to improve the resulting
motion of the robot. The environment analysis is to be
conducted in real-time.

'''1.2 Background'''

'''1.3 Objectives'''

(a)Design a software system for path planning to calculate and execute traversal to destination, while adopting an efficient path and avoiding collisions with obstacles.

(b)Design a hardware system, including a camera, computers, and a Quanser robot, to perform the task.

'''1.4 Project Structure'''

The entire project is divided into three major parts:

1.Environment Analysis

2.Path Planning

3.Motion Control

and the design of a graphical user interface (GUI) as the project team has three members. Michael Vieceli is responsible for the environment analysis, Yuanyang Wang focuses on path planning, and motion control is assigned to Di Zhang.
A GUI is designed to integrate individual’s work for hardware implementation.

== 2.System Overview ==

The figure below shows the integration of the whole system with input and output labeled.
[[File:]]

'''2.1 Hardware'''

2.1.1 Camera

2.1.2 Quanser

'''2.2 Software'''

2.2.1 Image processing software

2.2.2 Path planning software

2.2.3 Motion control software

== 3.Image Processing ==

== 4.Path Planning ==

'''4.1 Method'''

The intended method is based on A-star algorithm.
A-star algorithm is devised by P.E.Hart et al in 1968 [2] as an improved algorithm of the Dijkstra algorithm. It uses heuristic searching approach to calculate optimal path. The basic thought of this algorithm is to define a measurement standard, such as the length of path, then find out the waypoint with the least cost of all the possible waypoints. From this waypoint, waypoints tree is constructed and new waypoints with the least cost are addressed until the target is reached. The A-star algorithm is a heuristic search algorithm and works well in simplified environment.

'''4.2 Result'''

== 5.Motion Control ==
5.1 Kinematic Model

Differential drive often used in two-wheel drive system which each wheel is equipped with a separate actuator. The Quanser QBot 2 Mobile Platform uses it, two separately coaxial driven wheels placed on either side of the robot body which controlled by high-performance DC motors with encoders and drops sensors.
[[File:1.png|200px|thumb|bottom|Quanser QBot 2 Mobile Platformt]]
In the figure represent the radius of the wheels, the rotational speed of left and right wheel are wl and wr respectively. Thus, the equations of the linear speed of the left wheel and right wheel are vl=wl*r and vr=wr*r.
If only one of the two wheels rotates, the robot will rotate about the non-moving wheel. If two wheels rotate with the same speed, the robot will rotate one point which is far from the robot. The center of this point is Instantaneous Centre of Curvature (ICC). The distance from the center of the robot to the ICC is r(icc) and the distance between two wheels is d, θ is the angle of the robot, and the robot speed is vc. Then the equations of QBot 2 are:
[[File:234zd.png]]

Given the wheels speed are vl and vr , the robot speed is which all along the same axis. Then the following express the kinematic model for the different drive system:
[[File:2345zd.png]]

5.2 Trajectory tracking results

Control is the basement of navigation and path-planning of the robot, thus the higher requirement of stability, robustness, and accuracy are needed. As a further step of development to ensure correct operation of the robot, PID control using motion parameters and upcoming path steps to ensure ideal operation of the robot, with minimal overshoot and error inherent in the previous stages of motion design. To achieve desired controlling effect, the fuzzy PID method was employed to control the motions of tracking robot.
In the picture shows the simulation result of trajectory control.

[[File:23456zd.png|200px|Trajectory tracking results]]

== 6.Conclusion ==

== 7.Reference ==
[1] N Wang and P Chen, ‘Path planning algorithm of level set based on grid modeling’, in: Computer Design and Applications (ICCDA), International Conference, June 25-27, 2010.

[2] P. E. Hart et al., ‘A formal basis for the heuristic determination of minimum cost paths’, In: IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, no.2, July, 1968.

Projects:2016s2-250 On-Line Mains Power Cable Time Domain Reflectometry

2017-06-04T06:34:27Z

A1675729:

== Project Introduction ==

'''
A/Prof C. J. Kikkert has developed a power line communications (PLC) frequency impedance analyser for measuring impedance online
and presented this at a conference. Huawei would like to be able to measure distances between power-line
branches. This can be done using time-domain reflectometry. However normal time domain reflectometers cannot
be connected directly to the power mains while they are operating.
The project is to investigate techniques which can be used to measure reflections due to power-line branches. These
techniques include using time domain reflectometry with PLC frequency pulses or using radar type techniques with
the radar like pulses being transmitted on the power-line. Alternately the impedance versus frequency results
obtained by the on-line impedance analyser can be converted to impulse responses using the Fast Fourier Transform
and those can be used to determine reflections.
The outcomes from this project will be very useful to the power industry.
'''

== Project Team ==

=== Academic Supervisors ===

'''
A/Prof Keith Kikkert

A/Prof Nesimi Ertugrul'''

=== Team Members ===

'''
Jinyue Bai

Shuxin Tang
'''

== Headline text ==

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T06:11:54Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T06:01:19Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T06:00:51Z

A1675729: /* 2.System Overview */

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:56:13Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:55:35Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:53:02Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:41:33Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:41:11Z

A1675729:

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:36:45Z

A1675729: /* 2.System Overview */

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:36:36Z

A1675729: /* 2.System Overview */

Projects:2016s2-220 Path Planning and Collision Avoidance for Aduino Robots

2017-06-04T05:36:00Z

A1675729: