This paper conducts a simple study on the object detection system of the intelligent power inspection robot. Equipped with high-definition cameras, temperature sensors, buzzers, and other equipment, the intelligent obstacle avoidance patrol car can patrol along a preset path in the power plant and shoot at a fixed depression angle after reaching the designated location. Supplemented by MATLAB-based image processing algorithms, the robot could make prelimental judgments on the lines and electrical equipment. If the extracted images and environmental parameters exceed the preset safety threshold, or if there is a situation that cannot be independently identified, an alarm will be issued to alert the staff in the central control room. In this case, it will be judged by the staff or handled personally on the spot, and the data will be transmitted to the background through wireless transmission technology for inspection personnel to see.