A Vision-based Demonstration and Verification Platform for Path Planning

With the rapid development of robotics, path planning has become a hot issue. From the traditional artificial potential field method, genetic algorithm to the latest deep learning neural network, new path planning algorithms have emerged endlessly. However, there is little research on the construction of verification platform for path planning algorithm. In this paper, a visual based demonstration platform for path planning is designed. The information of scene is captured into the computer through a camera. This system combines perspective transformation and improved Canny detection to identify robots and obstacles, using the improved Hough circle transform and Kalman filter to improve the accuracy of the object position, getting the detailed information of the object in the scene. After using the path planning method to plan the path, the projection equipment will project it to the scene for visual demonstration. Finally, the artificial potential field method is demonstrated as an example.