Path Planning of Arbitrary Shaped Mobile Robots With Safety Consideration

This paper presents a neural network-based approach for the path planning of arbitrary shaped mobile robots in complex environments, with the consideration of safety. A 2D workspace is discretized to a topologically organized map using a biological neural network, in which the dynamic neural activity landscape represents the environmental information. A set of kernel matrices are established to describe the shape and orientation features of the robot. Taking the safety factor into consideration, the translation and rotation performances of the robot on each neuron node of the workspace are determined using a convolutional neural network (CNN). Then, from the initial state of the robot to the target state, a node rooted tree is constructed by searching the adjacent neurons, and the moving path of the robot is generated by backward searching the node rooted tree. By changing the bias coefficient in the convolutional calculation, the clearance between the planned path and the obstacles can be conveniently adjusted. The effectiveness of the proposed method is demonstrated through several simulations conducted in both static and dynamic environments. The results show that the method can effectively solve the ``path blocked'' issue caused by small densely scattered obstacles, and also solve the ''too close'' and ''too far'' path planning problems.