Research On Path Planning Of Intelligent Weeding Robot

Path planning technology is a fundamental function of weeding robot. The quantity of path planning algorithm has an important effect on intellectualization and efficiency.The current algorithms which still carry with some defects such as sub-optimality, lacking of instantaneity and colliding between roadblock and robot are merely concentrated in sectional aspects to give a resolution. There doesn’t exist one which is full of practicability. Aimed at the above issues this paper is focused on area-coverage path planning. The main content is characterized as follows:(1) The path planning method and area-coverage path planning method were lucubrated. A method combined global path planning and local path planning was propounded for weeding robot. The path planning of weeding robot which should be area-coverage path planning attached with maximum coverage rate and minimum repetition rate. The local path of intelligent weeding robot ought to be planned partially in order to avoid obstacle. As a consequence, edgewise studying and raster map method were put forward to establish model for working environment, similarly a global path planning algorithm with plowing style for global path planning and a neural network algorithm for local path planning.(2) A neural network based on cellular unit was proposed. A path planning algorithm based on recursion neural network that proposed by Yong Song was improved. The structure and algorithm of cellular neural network were analyzed at first. According to the non-uniqueness of the maximum activity location in the current vicinity a scheme was presented based on the drawback of recursion neural network path planning algorithm that proposed by Yong Song. The local path planning was modified when weeding robot stayed in the vicinity of obstacles. Meanwhile a simulation experiment was conducted for the sake of feasibility of algorithm.(3) The area-coverage path planning algorithm based on BP neural network was proposed, the same as a algorithm which cellular neural network and BP neural network were integrated for weeding robot. In consideration of the requirements of the area-coverage path planning, a local path planning was accomplished primarily,furthermore a simulation experiment was tested. After that the area-coverage path planning was put forward through the connection between advanced cellular neural network path planning algorithm and BP neural network algorithm. It can be concluded that both of the methods were effective and feasible according to the results from simulation test. While the latter can plan route dynamically which represents it has better adaptability to the environment and robustness.