Human-machine cooperation pipeline system based on deep reinforcement learning

According to a man-machine cooperation pipeline system based on deep reinforcement learning, through the deep learning technology, by means of RRBFNN, the purpose that a mechanical arm predicts the contact force applied by a human partner is achieved, an impedance controller of on-line self-tuning parameters is formed, the impedance model serves as a track planner to serve as an actor network of a reinforcement learning DDPG algorithm, and the real-time performance of the robot is improved. The task trajectory of the mechanical arm is optimized through a DDPG algorithm, and the efficiency-optimized man-machine cooperation assembly line system is achieved. Meanwhile, an SSD network is adopted to identify objects with different appearance characteristics, a Sobel operator and a Canny operator are adopted to form a complete object image edge, a fusion algorithm is proposed to fuse undetermined grabbing postures, and a final grabbing posture is formed to guide a mechanical arm to grab the objects. The capability