Automatic assembly line work efficiency optimization system and method based on reinforcement learning

The invention relates to an automatic assembly line work efficiency optimization system based on reinforcement learning. Another technical scheme of the invention provides an automatic assembly line work efficiency optimization method based on reinforcement learning. According to the system and the method, reinforcement learning is used in the carrying assembly line, and the carrying time of the whole assembly line is shortened through contact between the robot and the environment and continuous learning. On the basis of reinforcement learning, an OptLayer architecture is built; one-to-one simulation is carried out on a real object assembly line through a V-Rep platform; carrying out reinforcement learning by using an A3C algorithm; obtaining a training set on the simulation platform; thelearning time is greatly shortened; after a robot trains optimal parameters in a simulation environment, model parameters obtained after simulation training are migrated to a real object platform forself-learning training, an