Projects:2014S1-31 Autonomous Robot Navigation using a Movable Kinect 3D sensor

From Projects
Jump to: navigation, search

Project Introduction:

The goal of this project is to extend the functionality and navigation performance of an existing robot platform by adding a movable 3D Kinect sensor. The robot will be able to navigate through the environment and perform collision avoidance whilst it searches for an object of interest.

Objectives:

--Collision avoidance

--Object recognition

--Path planning


-31Robot.PNG

Team Members:

Yinzia Pang

Yingzheng Wang

Supervisors

Danny Gibbins

Braden Phillips


Overall Approach:

There are two parts to approach. One part is the Hardware and the other part is Software approach.

Hardware Approach:

The hardware structure is based on the previous project robot structure. The existing platforms are eight ultrasonic sensors, one Mux-shield board, one Wild-thumper board and four motors. To implement object recognition and path planning, a kinect sensor and a servo motor is added to the existing platform.


31Hardware.PNG

Software Approach:

The kinect image processing code and the Path Planning & Collision avoidance code are written in C++. Kinect image processing program not only output the object information, but also output wide range environment information. The output data then are used by the decision making program.


31Software.PNG


Object Recognition:

The Object Recognitoin is performed by 3D Kinect Sensor. In our project, the robot could recognize a simple object with a simple color. An example is showing below: A yellow ball is recognized.

31Object.PNG


Path planning:

In this project, the path planning mathod is 'Find the open space'. An example is showing below: the red line are suggesting the open spaces that detected

31Openspace.PNG


Collision avoidance:

Collision avoidance objective is performed by acoustic sensors. The logic of avoidance is showing below

31Avoidance.PNG


Testing:

A test result is showing below. It is suggesting that the robot is able to detect object, find the open space for navigation, and avoid collision.

31Test.PNG