METHOD AND APPARATUS FOR COLLISION AVIODANCE AND AUTONOMOUS SURVEILLANCE OF AUTONOMOUS MOBILE VEHICLE USING DEEP REINFORCEMENT LEARNING

摘要

According to an embodiment, a method and apparatus for generating a control parameter receives sensor inputs, and inputs a first state determined from a plurality of states corresponding to an arbitrary environment to the neural network according to the sensor inputs, thereby generating a first state Determines the values of the actions that can be executed in, determines the action corresponding to the value that maximizes the compensation in the first state among the values as the optimal action, and generates a control parameter for the device based on the optimal action.