Research On Task Planning Of Autonomous Rice Transplanter Based On Knowledge Graph

China is in a period of transition from mechanized agriculture to smart agriculture.As an important technology for transforming agricultural production models and realizing agricultural power,agricultural intelligent robots have gradually become the key research objects of various countries due to their advantages of efficient operation,safety,reliability,and independent intelligence.Task planning,as an important technology for agricultural intelligent robots to perform tasks autonomously,to a large extent marks the intelligent level of agricultural intelligent robots and is also a frontier hotspot in robot research.In this paper,research on task planning for autonomous operation rice transplanter is carried out.According to the needs of its operation and task planning,researches on task abstraction and expression,modeling of task planning system based on knowledge graph,design and analysis of task planning experiment are carried out.First,a knowledge graph oriented to the needs of robot operations is constructed.Aiming at the construction of knowledge graph,a robot task knowledge representation method based on triples is designed to extract and express the knowledge of robot operation,and construct a knowledge set.Then,according to the defined concept specification of robot task planning,the robot operation knowledge is imported into the database,and a knowledge graph for robot task planning is constructed.Then,a robot task planning method based on knowledge graph is designed.Search the knowledge graph through the search traversal strategy of depth-first and activity feature-matching priority to generate all task planning solutions;then analyze the relationship between the nodes in the initial solution,and optimize the overall planning by setting the target optimization strategy to obtain the optimal planning solution.The effectiveness of the method is verified by the comparison experiment of water supply task and the expansion experiment of robot task planning based on knowledge graph.Finally,for the independent operation of the rice transplanter,a knowledge map of the operation of the rice transplanter was constructed.Collected and organized the knowledge of rice transplanter operation and expressed it as a triple-structured knowledge collection,constructed the knowledge map of autonomous operation ricetransplanter,and carried out the simulation experiment of task planning of autonomous operation rice transplanter.All feasible sequences from the beginning to the end of the task,and can generate the optimal task planning plan,the result meets the actual job requirements.This article combines knowledge graph technology with robot task planning,and applies it to the task planning of the autonomous rice transplanter,which improves the robot’s ability to characterize and store task task knowledge,and can effectively form atomic task knowledge into contextual information that the robot can understand.It is expressed in the form of a network,which makes the robot task planning have a similar associative analysis ability as humans,and improves the robot’s intelligence level.