An alternating optimization approach integrating linkage design with motion planning for jumping robot. (July 2023)

• Record Type: Journal Article
• Authors: Gao, Haibo; Shan, Kaizheng; Wang, Shengjun; Han, Liangliang; Yao, Jie; Yu, Haitao
• Is Part Of: Mechanism and machine theory. Volume 185(2023)
• Publication Date: 2023-07
• Subjects: Alternating optimization -- Linkage design -- Motion planning -- Jumping robot; Machine theory -- Periodicals; Machinery -- Periodicals; Machines -- Périodiques; Génie mécanique -- Périodiques; Machine theory; Machinery; Periodicals; 621.81
• DOI: 10.1016/j.mechmachtheory.2023.105332
• Languages: English
• ISSNs: 0094-114X

Abstract

Highlights: A novel coaxial parallel five-link mechanism classification method is proposed. The leg linkage design problem is combined with motion planning of robot's body. An alternating optimization approach for jumping robot leg design is proposed. The experiments on biped robot show a jumping performance comparable to human. Abstract: Jumping, as widely observed amongst agile terrestrial creatures, is regarded as an indispensable ability for legged robots with high-performance in dealing with complex environments. This study presents a novel leg design methodology for jumping robot using the alternating iteration approach that optimizes the leg linkage and the specified jumping motion to completely leverage the capability of the actuator. First, a coaxial parallel five-link mechanism (CPFM) classification method was proposed to extract feasible candidates of leg configuration to alleviate the intensely searching complexity due to multi-link singularity. An alternating optimization paradigm was developed by elaborately combining the motion planning (MP) and the linkage design (LD) to generate a favorable leg with a CPFM morphology. The MP task was constructed to maximize the jumping distance of the robot while the LD task was proposed to minimize the peak demand of the actuators. Finally, a sagittal jumping robot prototype was developed based on the optimal results from the proposed paradigm. We obtained a horizontal jumping performance of 1.8 times the maximum leg length, which is comparable to the natural performance of a human.