Simultaneous Target Tracking And Obstacle Avoidance For Mobile Robot Using Kinect

With the development of science and technology, people has payed more and more attention on mobile robot technology. Now, we can find robot especially mobile robot from industry to daily life as well as from military application to space observation. In order to face the complex working environment, the robot is becoming more intelligent.The technology of mobile robot is a multidisciplinary technology, it contains kinematics control, environment information awareness, autonomous motion planning and so on. In order to design a mobile robot which can autonomous plan its path and track target in a complex environment, we use the image sensor Kinect as the main sensor to obtain the environment information. I combined the target tracking method with path planning and realized my algorithm in a differential-drive mobile robot that equipped a Kinect, the result confirmed that this system can successfully reach the control requirements.In this article, I first introduced the current research status of mobile robot and introduced a new3-D camera:Kinect. Kinect can gave us the color image of environment and the position of objects, and it is suitable for taking place of traditional ultrasonic and laser radar as the ranging tool.The RGB color image obtained by Kinect is used to tracking target and the particle filter algorithm based on color information is used. The tracking algorithm is sensitive and is liable to the distraction of the color-closed objects. So I combined the color information with the depth information and use them improving the particle filter algorithm’s weight calculating progress. The experiment result shows that this method can improve performance of tracking color-closed object.Artificial potential field algorithm is used for path planning which has clear physical significance and is easy to calculate. The path plan algorithm need the obstacles location obtained by Kinect, but the Kinect has a blind zone and when the object is too close, it will fail to get the object’s position. So we need to store the environment’s information as a map and update it to ensure the obstacle information’s integrity. In consideration of the inaccuracy of long distance location, this article use local map to store the environment information and use probability to determine whether there is an obstacle or not.There are many disadvantages when we use artificial potential field algorithm, many methods have been put forward to solve the destination unreachable and shocking problem. In this article, a potential method with safe region is put forward to improve mobile robot’s ability to get through the narrow path. This method takes the mobile robot’s moving direction into consideration and find a safe region. The experiment results confirm the algorithm is practicable.At last, I combine the target tracking with path planning, the experiment results show that Kinect can well take the place of traditional radar and also confirm the algorithms I put forward are practicable.