Operation method of reinforcement learning based agent using state memory based artificial neural network for collision avoidance and autonomous driving of autonomous vehicle and driving method of autonomous vehicle equipped with the agent

摘要

The present invention provides a method of operating a reinforcement learning-based agent using a state memory-based artificial neural network that enables autonomous moving objects to perform obstacle collision avoidance and autonomous driving even in indoor or outdoor environments where GPS shadow area or location recognition is not possible, and the agent Disclosed is a driving method of an autonomous mobile device equipped with