Robot path planning method and device based on HER-SAC algorithm

The invention discloses a robot path planning method and device based on an HER-SAC algorithm, and the method comprises the steps: building an environment and an obstacle in a physical engine, creating a target point, and initializing related parameters which comprise an award discount factor, an entropy parameter, a learning rate, a state space and an action space; an HER-SAC deep reinforcement learning network model is constructed, then the model is trained, and the training process is as follows: according to a target running robot, a running trajectory is sampled, trajectory experience is collected, and then the trajectory experience is stored in an experience playback pool; obtaining a new target by adopting a future sampling method, recalculating a new reward according to the state and action at the current moment, and putting the new reward into an experience pool; and selecting a sample from the experience pool, updating the value network and the strategy network, and then iteratively training the mod