基于Kinect的目标跟踪与避障

利用Kinect实现室内移动机器人的目标跟踪与避障功能,其中Kinect提供的光学图像用于实现机器人视觉目标跟踪。利用Kinect的人脸识别功能,避免当环境中出现多目标时,机器人对运动目标的识别出现混乱;同时增加了增量子空间对视频信息进行及时更新,避免机器人因目标移动较快而跟丢目标;利用T-S模糊神经网络法实现避障功能,相比传统算法有效改善了其收敛性。在移动机器人CRX10上实验证实了该系统能够较好地完成目标跟踪的同时具有躲避障碍物的功能。

Target Tracking and Obstacle Avoidance Using Kinect

By using the Kinect,we present a method for mobile-robot＇s target tracking and dynamic obstacle avoidance.The optical image provided by the Kinect is used to realize the robot＇s visual target tracking.We use the face recognition functionality of the Kiuect to avoid confusion in the tracking process,in case the robot can＇t identify the moving target correctly when multiple targets occurrs at the same time.We use an incremental subspace method to update the video information in a timely manner which will help the robot to seize the target accurately when the target moves fast to achieve obstacle avoidance,we use the T-S fuzzy neural network method which can improve effectively the convergence criterion than traditional algorithms.Finally,we experiment the system on a mobile robot CRX10 and validate the proposed algorithm,the result shows that it can realize target tracking and obstacle effectively.