In recent years, many wall-climbing robots have been developed in the field of petrochemical storage tank maintenance. However, it is difficult for most of them to be widely used due to common problems such as poor adsorption capacity, poor adaptation to elevation, and low trajectory tracking accuracy. In order to solve the problem of the robot not being able to achieve high-precision operation on curved surfaces, a new wall-climbing robot system is designed. Based on the magnetic wheel adsorption method, a passive adaptive motion mechanism that can adapt to walls with different curvatures is proposed. In order to improve the trajectory tracking accuracy of the wall-climbing robot, the kinematic model of the wall-climbing robot is simplified, a velocity compensation controller is designed, and the stability of the controller is proved by introducing the Lyapunov equation. Through experiments, the controller designed in this paper is compared with the conventional controller to verify the effectiveness and superiority of the controller. The experimental results show that the robot can move safely and stably on curved surfaces, with improved tracking accuracy and reduced trajectory deviation caused by response time lag, and meets the maintenance operation requirements of wall-climbing robots.

Rope-driven robots are increasingly being applied for the efficiently cleaning of glass curtain walls. However, increasingly complex wall surfaces and the various shapes of obstacles may block the robot and reduce coverage. In this study, three-DOF rope-driven cleaning robots and a full-coverage path-planning algorithm were developed to achieve global operation. The robot adopts a five-rope parallel configuration, and four winches are mounted on the wall and one on the ground to produce 3D motion performance. We used a grid method to build the wall model to mark obstacles, and then we decomposed it according to the wall curvature to better access cleaning subareas. To further increase the cleaning coverage rate, a full-coverage path-planning algorithm based on an improved priority heuristic was designed, which does not ignore the inset area of U-shaped obstacles. By introducing two sets of priority criteria to judge the forward direction, the robot can switch directions to cover a whole area when encountering U-shaped obstacles. Furthermore, by planning a return route requiring the least amount of time when entering a dead zone, an escape strategy was developed to prevent the robot from being unable to choose a direction. The experimental results show that the robot, after applying the proposed path-planning algorithm, could complete the global cleaning of complex glass walls with various obstacles.