Picking robot arm control method based on density clustering and agent integration
By integrating multiple agent algorithms through the DBSCAN algorithm and the error reciprocal weighted combination method of agent reward, the shortcomings of traditional harvesting robotic arms in motion control in complex agricultural environments are solved, and efficient and precise harvesting operations are achieved.
CN119458329BActive Publication Date: 2026-07-10HEBEI ELECTROMECHANICAL INTEGRATION PILOT BASE CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HEBEI ELECTROMECHANICAL INTEGRATION PILOT BASE CO LTD
- Filing Date
- 2024-11-14
- Publication Date
- 2026-07-10
AI Technical Summary
Technical Problem
Traditional harvesting robotic arms lack the precision and efficiency for motion control in complex and ever-changing agricultural environments, leading to crop damage and low efficiency.
Method used
The DBSCAN algorithm is used to cluster the spatial density of the harvested targets. Combined with the error reciprocal weighted combination method of agent reward, multiple agent algorithms are integrated to achieve precise and efficient control of the robotic arm.
Benefits of technology
The simplified harvesting environment model improved the efficiency and accuracy of the robotic arm, reduced crop damage, and increased harvesting efficiency.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN119458329B_ABST
Abstract
This invention belongs to the field of robotic arm motion planning and control technology, and discloses a control method for a harvesting robotic arm based on density clustering and agent integration. First, the DBSCAN algorithm is used to simplify the harvesting scenario, determining the objective function, constraints, input state space, action space, reward function, and termination condition of the agent in the harvesting scenario model. Based on the clustering results of the DBSCAN algorithm, the agent is prepared for training. Then, the selected agent algorithm is trained. Finally, an error reciprocal weighted combination method based on agent reward is used to integrate the motion control decisions of each trained agent algorithm for the same robotic arm and accumulate the total reward value after integration processing. This invention is applicable to the control of harvesting robotic arms, realizes multi-angle decision-making, effectively compensates for the shortcomings of single algorithms, improves the accuracy of robotic arm motion, and thus provides reliable support for modern agricultural automation technology.
Need to check novelty before this filing date? Find Prior Art