Robot state planning method based on Monte Carlo tree search algorithm

US12370676B2 · kind B2 · utility

Assignee

• HARBIN INSTITUTE OF TECHNOLOGY · CN

Inventors

• Haibo Gao · Jingwei, CN
• Liang Ding · Zhejiang, CN
• Peng Xu · Beijing, CN
• Yan Liu · Elmsford, US
• Zhikai Wang · Xing'an, CN
• Zongquan Deng · 双山街道, CN
• Zhen-Xing Liu · Nanhu, CN

Key dates

• Filing date: Sep 24, 2020
• Grant date: Jul 29, 2025
• Expiry date: Jun 26, 2041

Classification

• Technology area (CPC B): Performing Operations; Transporting
• CPC primary: B25J9/1664
• WIPO field: Handling
• WIPO sector: Mechanical engineering

Abstract

A robot state planning method based on a Monte Carlo tree search algorithm. The method comprises: obtaining an initial state and a target state of a robot (step 100); by using the initial state as a starting node, using a Monte Carlo tree search algorithm to expand a Monte Carlo tree until a generated target node reaches the target state (step 200); and determining a state sequence of the robot according to all nodes from the starting node to the target node (step 300). According to the solution, the overall state in the motion process is planned, and a state sequence is generated; thus, effect of front back coupling caused by planning according to periods can be avoided, and the passing capability of a six-foot robot in a complex terrain is improved.