AGV path planning based on deep reinforcement learning

Abstract

The invention discloses AGV path planning based on deep reinforcement learning. According to the AGV path planning based on deep reinforcement learning, a simulation environment and an AGV model are built according to a ros operating system; designing the state, action and reward of reinforcement learning, and building an algorithm framework of deep reinforcement learning; training the network, and enabling the AGV to learn a strategy from the environment; and the trained network is used to realize AGV path planning. According to the method, the optimal action is acquired from the original image acquired from the AGV camera, and the features are automatically extracted; the problem that a traditional algorithm is prone to falling into a local minimum value is solved. No priori knowledge of the environment is needed, the environment is explored by itself, and feasible strategies are learned. When the environment is changed in real time or the environment condition is complex, dynamic changes can be made to avoid dynamic obstacles, and a target point is reached.

Description

technical field [0001] The invention belongs to the field of artificial intelligence and relates to AGV path planning based on deep reinforcement learning. Background technique [0002] Since the 21st century, with the rapid development of science and technology, artificial intelligence technology has become the cutting-edge of science and technology. It is one of the important research topics of scientific research institutions, technology companies and major universities in various countries, and it also represents an important direction of future scientific and technological development. AGV (Automated Guided Vehicle) is called an automatic guided vehicle. In the field of industrial manufacturing, industrial AGV has become an important productivity. In addition to the field of industrial manufacturing, AGV also appears in all aspects of people's lives, from large shopping malls to parks to large hospitals, you can find the shadow of AGV. During the epidemic, the non-con...

AI Technical Summary

Problems solved by technology

However, in more complex dynamic path planning problems, AGV needs to face more uncertainties in the environment. Traditional algorithms are easily trapped in local minimums, and may even fail to plan a feasible path.

Images