A New Wave Neural Network Dynamics for Planning Safe Paths of Autonomous Objects in a Dynamically Changing World

We consider the problem of finding a safe path for a robot/manipulator in a dynamic environment and propose a novel neural network model for solving this task. The network has discrete time-dynamics, is locallyconnected, and is, hence, computationally efficient. No preliminary information about the current world status is required for the planning process. Path generation is performed via the neural-activity landscape, which forms a dynamically-updating potential field over a distributed representation of the configuration space of an object. The network dynamics guarantees local adaptations, and includes a set of strict rules for determining the next step in the path of an object. According to these rules, planned paths tend to be optimal in a L1 metric. We present here the description of the model, and evaluate simulation results for various types of environmental changes. Key-Words: path planning, neural networks, wave expansion, robotics, autonomous navigation