Dynamic obstacle avoidance for a pioneer robot
Khoo, Victor Boo Soon
Date of Issue2016-05-13
School of Electrical and Electronic Engineering
This project aims to design a obstacle detection and avoidance module for a Pioneer robot using low cost sensor so that it can detect and avoid dynamic obstacles intelligently. The dynamic obstacles include moving human beings, robots, and vehicles. Last but not least, evaluate the effectiveness of detection and obstacle avoidance algorithms through experimental tests on a Pioneer robot. Most existing projects uses laser range finder as their sensor in their obstacle detection application. Laser range finder has many advantages and offer great range of detection up to 30metres. However, it is expensive to owe a laser range finder and does not fit with the project requirement of using low cost sensor. On 4th November 2010, Microsoft released the Xbox Kinect sensor that uses a RGB camera and a 3D depth ranging sensor for gaming application. The kinect is nearly 1/10 of the price of a laser range finder but yet offer a cost effective solution using a infrared light to generate depth pixel, adding a 3rd dimension to the image received. However the kinect does have its own limitation. For detection, by utilizing an inexpensive kinect sensor, it provides the depth information of the environment and obstacles for the pioneer robot. From the depth information, the program takes the sampling point cloud in real time to process for an autonomous navigation with obstacle avoidance. One important information of the depth information is motion parallax. Motion parallax is a depth cue related to the dynamic of the obstacle. For avoidance, the focus is on the effectiveness of the algorithm on how to avoid the obstacles safely in a indoor environment. A effective obstacle avoidance scheme should be optimal with respect the desired speed and kinematics of the robot, the quality information provided by the sensors as well as to determine of future risk of collision. The pioneer 3-AT work effectively in static obstacle avoidance however there are some issues with dynamics obstacle avoidance which will be documented in this thesis.
Final Year Project (FYP)
Nanyang Technological University