Auto-Optimizing Connection Planning Method for Chain-Type Modular Self-Reconfiguration Robots

被引:4
|
作者
Luo, Haobo [1 ,2 ]
Lam, Tin Lun [1 ,2 ]
机构
[1] Chinese Univ Hong Kong, Sch Sci & Engn, Shenzhen, Guangdong, Peoples R China
[2] Chinese Univ Hong Kong, Shenzhen Inst Artificial Intelligence, Robot Soc AIRS, Shenzhen 518172, Peoples R China
基金
中国国家自然科学基金;
关键词
Computational complexity; connection planning; graph matching; modular robots; self-reconfiguration; CONFIGURATION RECOGNITION; DISTRIBUTED CONTROL; LOCOMOTION; ALGORITHMS;
D O I
10.1109/TRO.2022.3218992
中图分类号
TP24 [机器人技术];
学科分类号
080202 ; 1405 ;
摘要
Chain-type modular robots are capable of self-reconfiguration (SR), where the connection relationship between modules is changed according to the environment and tasks. This article focuses on the connection planning of SR based on multiple in-degree single out-degree (MISO) modules. The goal is to calculate the optimal connection planning solution: the sequence with the fewest detachment and attachment actions. To this end, we propose an auto-optimizing connection planning method that contains a polynomial-time algorithm to calculate near-optimal solutions and an exponential-time algorithm to further optimize the solutions automatically when some CPUs are idle. The method combines rapidity and optimality in the face of an NP-complete problem by using configuration pointers, strings that uniquely specify the robot's configuration. Our polynomial-time algorithm, in-degree matching (IM) uses the interchangeability of connection points to reduce reconfiguration steps. Our exponential-time algorithm, tree-based branch and bound (TBB) further optimizes the solutions to the optimum by a new branching strategy and stage cost. In the experiments, we verify the feasibility of the auto-optimizing method combining IM and TBB, and demonstrate the superiority of IM over Greedy-CM in the SR of MISO modules and the near-optimality of IM compared to the optimal solutions of TBB.
引用
收藏
页码:1353 / 1372
页数:20
相关论文
empty
未找到相关数据