analiza aiqbench
The paper's core contribution is an RL method: a population of specialist policies that co-designs robot morphology and control, aimed at a known failure mode (behavioral diversity collapse) in shared policies. The reported gains are 5-9 cm in obstacle height and about 67% less wall-clock co-design time, though these come from one robot (BALLU) on a few tasks with up to four design dimensions. Readers should check whether the approach generalizes beyond that setting, since the hardware validation is teleoperated and limited to a single optimized morphology.
Rapid co-design of Buoyancy-assisted robots for Challenging Locomotion using Gaussian Evolutionary Specialists
arXiv:2606.07424v2 Announce Type: replace Abstract: Designing high-performance legged robots requires jointly optimizing morphology and control. Model-free reinforcement learning (RL) enables learning complex controllers during co-design, but repeatedly training policies across candidate morphologies makes design optimization computationally expensive. Existing approaches can amortize policy learning across morphologies, but a single shared policy tends to converge toward one control strategy regardless of morphology, which we refer to as behavioral diversity collapse. We introduce Gaussian Evolutionary Specialists (GES), an RL-based co-design framework that maintains a population of specialist policies whose design-space territories evolve according to their relative control performance. We evaluate GES on the Buoyancy-Assisted Light Legged Unit (BALLU) robot for obstacle traversal and ramp walking over design spaces of up to four dimensions. GES achieves higher best design performance than Monolithic and Mixture-of-Experts baselines, with gains of 5.0-9.0 cm in obstacle height and 10.0-12.7 points in ramp grade. For the same 200 design-optimization evaluations, GES reduces wall-clock co-design time by 67.4% compared with retraining a policy for every candidate. Finally, teleoperated hardware experiments evaluate the optimized BALLU morphology, which traverses obstacles of up to 24 cm, compared with 8 cm on the original design.
Abstract:Designing high-performance legged robots requires jointly optimizing morphology and control. Model-free reinforcement learning (RL) enables learning complex controllers during co-design, but repeatedly training policies across candidate morphologies makes design optimization computationally expensive. Existing approaches can amortize policy learning across morphologies, but a single shared policy tends to converge toward one control strategy regardless of morphology, which we refer to as behavioral diversity collapse. We introduce Gaussian Evolutionary Specialists (GES), an RL-based co-design framework that maintains a population of specialist policies whose design-space territories evolve according to their relative control performance. We evaluate GES on the Buoyancy-Assisted Light Legged Unit (BALLU) robot for obstacle traversal and ramp walking over design spaces of up to four dimensions. GES achieves higher best design performance than Monolithic and Mixture-of-Experts baselines, with gains of 5.0-9.0 cm in obstacle height and 10.0-12.7 points in ramp grade. For the same 200 design-optimization evaluations, GES reduces wall-clock co-design time by 67.4% compared with retraining a policy for every candidate. Finally, teleoperated hardware experiments evaluate the optimized BALLU morphology, which traverses obstacles of up to 24 cm, compared with 8 cm on the original design.
Robotics (cs.RO)
arXiv:2606.07424 [cs.RO]
(or arXiv:2606.07424v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.07424 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 277 cuvinte