Technology Research, Vision-based Wheeled Mobile Robot Target Tracking

For Wheeled Mobile Robot(WMR), moving by getting information offered from vision is one of the hottest research in the world. This paper does a lot of work deeply in detail about locating and tracking moving object under the circumstances when both the robot and the camera set upon the robot could move.This paper has designed the whole frame of the task by using modularized design which sets the whole task apart. Therefore, the whole task includes building and starting system block, image process block and strategy block. These blocks cooperate together to finish the goal of object recognition, locating and tracking the moving target.In image process block, this paper takes the preprocess method fit for the task. In image segmentation, this paper proposes thresholding method which has been changed a bit to make it more suitable for the task. The experiment result has proved that this method could segment the image to different areas. In target recognition, this paper takes area based-invariant moments method according the shape characteristic, besides, the paper use Kalman Filter method to predict position and velocity of the interest point successfully and the experiment demonstrates that Kalman Filter methos is exact and stable. In addition, the design of changing the range of image process according to situation is of help for reducing calculating time and object recognition error .In moving strategy, this paper use PID method to control wheel speed by reading counter. For the need of changing co-ordinate system, this paper calculates the formula of camera theory. For the need of control robot and camera to work together, this paper calculates the relationship of the object change on image and turning angle of the camera. And this paper designs the method of controlling the robot and camera work together.This paper implements the module function under Visual C++ integrated develop environment. The experiment results prove that the robot could recognize the target precisely in set time and track the target.