Research On Path Planning For Water Surface Garbage Cleaning Robot

With the development of China’s comprehensive national strength,the health and safety of water bodies have received widespread attention from researchers and the public.President Xi Jinping proposed that "clear waters and green mountains are as valuable as mountains of gold and silver".Floating debris on the water surface can cause some damage to the organisms in the water body and need to be salvaged in order to protect the water safety and the life of the organisms in the water.However,existing water surface salvage devices are not suitable for small water quality,design a surface garbage removal robot.In order to improve the efficiency of the robot,this article studies the path planning and target recognition of the water surface garbage cleaning robot.The main research contents are as follows:Firstly,conduct kinematic analysis on the body of the water surface garbage cleaning robot and the robotic arm of the garbage collection device,and validate the model of the water surface garbage cleaning robot on a simulation software platform,providing a reference basis for subsequent research.Secondly,to improve the efficiency of the robot and plan a safe path,the path planning of the water surface garbage cleaning robot is studied,and a collision-free path from the starting point to the ending point is planned using a genetic algorithm.Due to the premature convergence and local optimum problems of genetic algorithms,the genetic algorithm is improved by integrating simulated annealing algorithm into the crossover operation of the genetic algorithm.A grid map that matches the environment is established,and comparative experiments of the algorithms are carried out in Matlab simulation software.It is found that the genetic-simulated annealing algorithm can avoid premature convergence and local optimum problems,and improve the feasibility and reliability of the algorithm.Then,deep learning-based target recognition algorithms are highly robust in water scenarios compared to traditional target recognition algorithms.In this thesis,the Yolov5 object recognition algorithm is used to detect floating objects on the water surface.It is trained and tested on a professional dataset,and the Yolov5 algorithm is improved by changing the initial anchor box,adding attention mechanism,optimizing the loss function,etc.The improved Yolov5 algorithm improves precision and average precision by 3.8%and 0.8%,respectively.Finally,the control system design scheme of the water surface garbage cleaning robot is introduced,including hardware and software.Sunrise PI ARM embedded development board is used as the core board of robot,and STM32F103C8T6 is selected as the main control chip of the driver board.The peripheral circuit of the main control chip is introduced based on the implemented functions.The software mainly introduces the operational flow and image recognition,and verifies the path planning and object recognition of the robot on the experimental platform.