Fast Algorithms to Test Robust Static Equilibrium for Legged Robots - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2016

Fast Algorithms to Test Robust Static Equilibrium for Legged Robots

Résumé

Maintaining equilibrium is of primary importance for legged systems. It is not surprising then that static equilibrium is at the core of most control/planning algorithms for legged robots. Being able to check whether a system is in static equilibrium is thus important, and doing it efficiently is crucial. While this is straightforward for a system in contact with a flat ground only, it is not the case for arbitrary contact geometries. In this paper we propose two new techniques to test static equilibrium and we show that they are computationally faster than all other existing methods. Moreover, we address the issue of robustness to errors in the contact-force tracking, which could lead to slippage or rotation at the contacts. We extend all the discussed techniques to test for robust static equilibrium, that is the ability to maintain equilibrium while avoiding to lose contacts despite bounded force-tracking errors. Accounting for robustness does not affect the computation time of the equilibrium tests, while it qualitatively improves the contact postures generated by our reachability-based multi-contact planner.
Fichier principal
Vignette du fichier
final_submission.pdf (2.45 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01201060 , version 1 (17-09-2015)
hal-01201060 , version 2 (25-02-2016)

Licence

Paternité - Partage selon les Conditions Initiales

Identifiants

  • HAL Id : hal-01201060 , version 2

Citer

Andrea del Prete, Steve Tonneau, Nicolas Mansard. Fast Algorithms to Test Robust Static Equilibrium for Legged Robots. IEEE International Conference on Robotics and Automation (ICRA), May 2016, Stockholm, Sweden. pp. 1601-1607. ⟨hal-01201060v2⟩
280 Consultations
701 Téléchargements

Partager

Gmail Facebook X LinkedIn More