Neural network approaches to dynamic collision-free trajectory generation

In this paper, dynamic collision-free trajectory generation in a nonstation ary environment is studied using biologically inspired neural network appro aches, The proposed neural network is topologically organized, where the dy namics of each neuron is characterized by a shunting equation or an additiv e equation. The state space of the neural network can be either the Cartesi an workspace or the joint space of multi-joint robot manipulators, There ar e only local lateral connections among neurons. The real-time optimal traje ctory is generated through the dynamic activity landscape of the neural net work without explicitly searching over the free space nor the collision pat hs, without explicitly optimizing any global cost functions, without any pr ior knowledge of the dynamic environment, and without any learning procedur es. Therefore the model algorithm is computationally efficient, The stabili ty of the neural network system is guaranteed by the existence of a Lyapuno v function candidate. In addition, this model is not very sensitive to the model parameters, Several model variations are presented and the difference s are discussed. As examples, the proposed models are applied to generate c ollision-free trajectories for a mobile robot to solve a maze-type of probl em, to avoid concave U-shaped obstacles, to track a moving target and at th e same to avoid varying obstacles, and to generate a trajectory for a two-l ink planar robot with two targets. The effectiveness and efficiency of the proposed approaches are demonstrated through simulation and comparison stud ies.