An Efficient Architecture for Robotic Path Planning

There are many path planning algorithms designed for mobile robots with software implementation. In the case of dynamic environments high-speed planning and recomputation of paths is necessary to avoid collision of robots with moving objects. A hardware-efficient algorithm is presented for finding a path of a mobile robot on image of an environment captured by an overhead camera. The algorithm computes the shortest path to identify collision free region for the robot. If multiple paths are provided, the path is traced to nearest path. 2-D cell architecture has been presented for the Euclidean distance Transform (EDT) and Nearest Neighbor Transform (NNT) for operating at high speed. The simulation result shows that the architecture is suitable for path planning in a dynamic environment containing obstacles with arbitrary shape and motion.