Reconfigurable legged metamachines that run on autonomous modular legs
🤖 AI Summary
Current legged robots lack bio-inspired reconfigurability, hindering adaptation to novel tasks or recovery from physical damage, and are constrained by fixed quadrupedal or bipedal morphologies. To address this, we propose a reconfigurable legged hyper-robot system comprising autonomous, modular, single-degree-of-freedom, high-dynamic leg units enabling plug-and-play physical and communication integration. We introduce a novel “design genome”—a latent-space encoding scheme—that efficiently generates and optimizes vast numbers of atypical legged configurations. Coupled with distributed reinforcement learning control, the system achieves meter-scale high-speed dynamic locomotion, robust traversal of unstructured terrain, and sustained operation under severe structural damage. Experimental validation demonstrates over thirty previously unexplored legged morphologies, breaking conventional configurational paradigms. This work establishes a new framework for embodied agents capable of adaptive morphological evolution.
Application Category
📝 Abstract
Legged machines are becoming increasingly agile and adaptive but they have so far lacked the basic reconfigurability of legged animals, which have been rearranged and reshaped to fill millions of niches. Unlike their biological counterparts, legged machines have largely converged over the past decade to canonical quadrupedal and bipedal architectures that cannot be easily reconfigured to meet new tasks or recover from injury. Here we introduce autonomous modular legs: agile yet minimal, single-degree-of-freedom jointed links that can learn complex dynamic behaviors and may be freely attached to form legged metamachines at the meter scale. This enables rapid repair, redesign, and recombination of highly-dynamic modular agents that move quickly and acrobatically (non-quasistatically) through unstructured environments. Because each module is itself a complete agent, legged metamachines are able to sustain deep structural damage that would completely disable other legged robots. We also show how to encode the vast space of possible body configurations into a compact latent design genome that can be efficiently explored, revealing a wide diversity of novel legged forms.
Problem
Research questions and friction points this paper is trying to address.
Lack reconfigurability in legged machines compared to animals
Limited adaptability of current quadrupedal and bipedal robot architectures
Difficulty in rapid repair and redesign for dynamic environments
Innovation
Methods, ideas, or system contributions that make the work stand out.
Autonomous modular legs enable reconfigurable legged metamachines
Compact latent design genome explores diverse body configurations
Single-degree-of-freedom jointed links learn dynamic behaviors
🔎 Similar Papers
No similar papers found.
