Path Planning Method for Substation Protective Barriers Deploy Robot Based on Deep Learning

Deploying electrical protective barriers to build isolation exclusion zones is an important prerequisite for substation power equipment maintenance operations. Automatic deployment of protective barriers by machines instead of humans has become a field requirement but faces challenges. To solve the problem of optimizing the path for a robot to transport and deploy protective barriers to the area, and to form a restricted area when global information is missing, an adaptive hierarchical path planning method is proposed based on deep reinforcement learning. Considering the lack of global information in the station, a Markov decision chain is used to establish a local information interaction navigation model for the robot transporting and deploying the protective barriers to the area. Considering the spatial geometry and load constraints of the transporting protective barriers, the dynamic orientation of the robot in the feasible path is optimized based on the fusion of multi-source sensing information. The scenario map in which the robot deploys the fence to form an exclusion zone is rasterized, and a dynamic interactive obstacle avoidance is achieved for the fence deployment path in the raster network. Dynamic interaction obstacle avoidance, using deep Q-network algorithm, extracting the ultra-wideband position information of robot instant positioning to generate composite state features, introducing three elements of obstacle avoidance equipment distance, obstacle space safety boundary and collision reward and punishment mechanism in the reward function, to achieve adaptive obstacle avoidance path planning under the task of robots laying loaded fences to form forbidden zones in a sequential manner. Simulations and experiments verify the effectiveness of the proposed method. The proposed method for time-sharing path navigation in the task of laying the exclusion fence improves the success rate of the navigation of the substation fence placement robot, and lays the foundation for the fully automated laying of the protective barriers technology.