Projects:2018s1-170 Intelligent Parking Control for Autonomous Ground Vehicles

From Projects
Revision as of 23:41, 8 April 2018 by A1709295 (talk | contribs) (IV. Members)
Jump to: navigation, search

I. Introduction

I.I. Background

In recent years, with the rapid increasing in the number of private cars, parking problems in large and medium-sized cities have become more and more serious. People in daily life have been accustomed to handing over problems to artificial intelligence. Therefore, there is an urgent need to design an intelligent parking system. The intelligent parking system integrates computer networks, video surveillance, image recognition and processing, and automatic control technologies to manage the vehicles automatically in the parking lot. Compared to the traditional parking lot, the intelligent parking system can accurately distinguish in-system vehicles and exotic vehicles. It can also save management personnel expenses, improve work efficiency and economic efficiency. Moreover, the smart parking system has good scalability and it can be applied to various unmanned scenarios.

I.II. Objective

Intelligent Parking Control for Autonomous Ground Vehicles project involves in designing a multi-vehicle system which includes 4 robots for autonomous location, guidance, and parking in a congestion parking area. Each robot represents an individual ground vehicle which can independently select best visual parking lot while avoiding visible obstacles. Obstacles avoidance not only including visual still obstructions, but also including moving visual obstacles. Camera installed to track the velocity of moving obstacles, and algorithm would be used for vehicles to avoid those obstacles.

I.III. Motivation

This project is motivated by simplifying the difficult and complicated parking process when driving a vehicle and improving the efficiency in group parking control. This system is aimed for multi-platform, future work may extend it to UAVs, UUVs and some other kind of unmanned applications.

II. Literature review

II.I. Environmental data detection

Environmental data detection is the use of computer algorithms to perform image processing on digital images and analyze the resulting environmental data. Since the image is defined in two dimensions (possibly more), environmental data detection may be modeled in a multidimensional system (Ramya R, Anand Kumar S, Krinish N K, Suraj V, 2015). The goal of environmental data detection is to improve the visualization of locations of targets by optimizing these physical parameters. The processing parameters need to be chosen correctly to overcome the opposite relationship between contrast and latitude, while at the same time getting the exact position of the desired target.

II.II. Path planning

For the path planning of a single robot, we only need to consider the shortest distance from the starting point to the destination. According to Dijkstra's algorithm, we can divide the map captured by the camera into several squares. By calculating the heuristic equation of F = G + H for each square, the path with the shortest distance could be determined. It means the smaller value of F, the shorter distance the robot will travel. When considering multiple robots, we first need to find the shortest distance from each robot to different parking spaces according to Dijkstra's algorithm and then find the optimal solution for multiple robots' selection of parking spaces in terms of Hungarian algorithm. In addition, since the Hungarian algorithm is a method that can avoid the conflict between robots, we do not need to consider the case that multiple robots select the same parking space at the same time.

II.III. Virtual field force

Ali Marzoughi (2017) assumed a methodology for robots to avoid obstacles. Regarding each visual obstacle is doing a repulsive force to a pointwise vehicle. At the meantime, the destination is doing an attractive force to the vehicle. Velocity of the moving vehicle could be influenced by these two kinds of force. Magnitude of each force is changed in real time regarding the velocity and distance of the obstacles and the destination.

II.IV. Motion support

In the execution part, QBot2 can integrate with MATLAB/Simulink software to provide real-time communication and interfacing to the components of the QBot 2. According to a typical kinematics model that computes the robot chassis speed Vc, turning rate ω, the wheel speed Vr and Vr, the wheel separation distance d to calculate the x and y-axis coordinates, and to control and move the QBot 2 in the scheduled path. For execution, the first thing needs to do is to find path and location based on image features and obstacle avoidance algorithm. This information, along with the location and orientation of the robot chassis, can be used for autonomous map building. Next, we should sign motion planning involves the creation of motion commands for the QBot 2 based on a series of goal positions or way-points.

III. Abstract

Intelligent Parking Control for Autonomous Ground Vehicles project involves in designing a multi-vehicle system which includes 4 robots. Robots in this system are QBOT2 and are integrated with camera to detect environmental information. Each robot in the system is individual and able to independently choose its parking space and avoid collision with obstacles and other vehicles. The motivation of this project is to provide a simplified, efficient parking environment. In addition, future work may extend it to UAVs, UUVs and some other kind of unmanned applications.

IV. Members

Chenyuan Wang;

Chengcheng Mao;

Dahai Hu;

Junxi Liu;

V. Supervisors

Prof. Peng Shi;

Prof. Cheng-Chew Lim;

VI. Advisors

Yutong Liu;

Bing Yan;

Xin Yuan

VII. References

Quanser Inc., QBOT 2 Workbook – Student, 2015

Quanser Inc.,QBOT 2 - User Manual, 2015

Ercan Taskiran , Yilmaz Durna and Hasan Kocer, “Wi-Fi Control of Mobile Robot Motion Types Based on Differential Drive Kinematics Modelling Approach” in Advanced Technology & Science, April 2016, pp 170-173

Yu, H Y 2010, ‘Multi-model distribution problem based on Hungarian algorithm’, Delivery Technology, vol. 29, no. 11, pp. 74-75.

The Qbot-2-Quarc 2018, digital photograph, Quanser, viewed 1 March 2018,<https://www.quanser.com/products/qbot-2-quarc>.

The algorithm of Path Planning 2017, CSDN, Technology Blog, viewed 13 March 2018,<https://blog.csdn.net/u012907049/article/details/78037686>.

Path Planning of Robots 2017, CSDN, Technology Blog, viewed 14 March 2018,<https://blog.csdn.net/darren2015zdc/article/details/73495441>.