The Design Of Vision System Of The Wall-cleaning-climbing Robot

In recent years,the wall-climbing-cleaning robot has gradually become the focus of research in the shipbuilding industry because it is more environmentally friendly,efficient and safe than manual sandblasting.At present,wall-climbing-cleaning robots at home and abroad mostly use remote control.The cleaning trajectory during the cleaning process is difficult to control,the cleaning efficiency is low,and workers are easily fatigued.The subject of the vision system for the wall-cleaning-climbing robot is supported by the project ’The development and application of complete equipment of marine derusting robot’ based within the Support of Science and Technology Planning Project of Guangdong Province,aiming at developing a more intelligent and automatic climbing wall cleaning robot,and promote its use in the shipbuilding industry.In this paper,a vision system for a wall-climbing-cleaning robot surface derusting detection and path planning is proposed.It uses Open CV to design visual recognition algorithms,and uses QT for modular system architecture.The functional modules of the robot vision system including hardware selection,image pre-processing,data post-processing,composition of the system architecture and description of the working principle are given in this paper.According to the different working conditions of the robot,a robot pose model was established.According to the overall design of the system operation process,the entire system is divided into three functional modules.Multi-thread,the signals and slots mechanism is used to make each functional module as a whole.Finally,a wall-climbing-cleaning robot vision system was designed,and an experimental platform was set up for performance experiments,the detection accuracy of the preprocessed image is 2.7mm for horizontal static accuracy,and the measurement accuracy for attitude tilt is 0.9°.The accuracy and reliability of the visual recognition system were verified through the experimental platform built and the actual tests in the shipyard.The experimental results show that the vision system can complete the rust spot recognition and path detection by the hull collected images,which meets the actual industrial requirements and reaches the research goal of this paper.The vision system solves the cleaning inefficiency problem of the cleaning robot and makes the hull cleaning robot more automated.