通过可变减速比设计,提升人形机器人膝关节爆发力。
Explosive Output to Enhance Jumping Ability: A Variable Reduction Ratio Design Paradigm for Humanoid Robots Knee Joint
- 膝关节采用渐变减速比设计,启动时高减速比增大力矩。
- 实验实现63厘米垂直跳,比最优固定比提升28.1%。
- 适合追求高机动性的仿人机器人跳跃性能优化者。
提升人形机器人膝关节的爆发功率对增强敏捷性和越障能力至关重要。然而,膝关节与质心间传动比与跳跃需求不匹配,加之高速下电机性能下降,限制了高功率输出持续时间,制约跳跃表现。本文提出一种新型膝关节设计范式,利用跳跃过程中动态减小的减速比实现爆发输出。基于电机输出特性与膝关节运动学分析,设计出随关节伸展逐渐降低减速比的耦合策略:初始高减速比快速提升起跳扭矩,逐步减小则抑制电机转速上升与功率损失,从而维持持续高功率输出。采用紧凑高效的直线电机驱动导杆机构实现该策略,并通过爆发跳跃控制策略指导参数优化。实验验证显示,在单关节平台上实现63厘米垂直跳跃(理论较最优固定减速比提升28.1%)。集成至人形机器人后,实现1.1米远跳、0.5米垂直跳及0.5米箱跳。
原文摘要 · Abstract (English)
Enhancing the explosive power output of the knee joints is critical for improving the agility and obstacle-crossing capabilities of humanoid robots. However, a mismatch between the knee-to-center-of-mass (CoM) transmission ratio and jumping demands, coupled with motor performance degradation at high speeds, restricts the duration of high-power output and limits jump performance. To address these problems, this paper introduces a novel knee joint design paradigm employing a dynamically decreasing reduction ratio for explosive output during jump. Analysis of motor output characteristics and knee kinematics during jumping inspired a coupling strategy in which the reduction ratio gradually decreases as the joint extends. A high initial ratio rapidly increases torque at jump initiation, while its gradual reduction minimizes motor speed increments and power losses, thereby maintaining sustained high-power output. A compact and efficient linear actuator-driven guide-rod mechanism realizes this coupling strategy, supported by parameter optimization guided by explosive jump control strategies. Experimental validation demonstrated a 63 cm vertical jump on a single-joint platform (a theoretical improvement of 28.1\% over the optimal fixed-ratio joints). Integrated into a humanoid robot, the proposed design enabled a 1.1 m long jump, a 0.5 m vertical jump, and a 0.5 m box jump.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品;中文卡片由大模型生成,请以原文为准。