(BIBTEX) Fernbach18

CROC: Convex Resolution Of Centroidal dynamics trajectories to provide a feasibility criterion for the multi contact planning problem. Pierre Fernbach, Steve Tonneau and Michel Taïx. In 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems, October 2018. (PDF)

Abstract

 We present a novel method for computing centroidal dynamic trajectories in multi-contact planning context. With dynamic motion it is necessary to respect kinematic and dynamic constraints during the contact planning step. Verifying the feasibility of a transition between contacts increase the success rate of the motion generation along the planned contacts. Our approach is based on a conservative but convex reformulation of the problem where we represent the center of mass trajectory as a Bezier curve, with control points constrained by the initial and final states and one free control point. Thanks to the convexity of this formulation, we can solve it efficiently with a Linear Program of low dimension. We use this LP as a feasibility criterion to test the contact transition candidates during multi-contact planning. By incorporating this criterion in an existing sampling-based contact planner, we are able to produce more robust contact sequences. We illustrate this application on various multi-contact scenarios. We also show that we can compute valuable initial guess, used to warm-start non-linear solvers for motion generation methods. This method could also be used for the 0 and 1-Step capturability problem. 

Pdf

Bibtex entry

@INPROCEEDINGS { Fernbach18,
    TITLE = { {CROC: Convex Resolution Of Centroidal dynamics trajectories to provide a feasibility criterion for the multi contact planning problem} },
    AUTHOR = { Fernbach, Pierre and Tonneau, Steve and Ta\"ix, Michel },
    BOOKTITLE = { 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems },
    URL = { https://hal.archives-ouvertes.fr/hal-01726155 },
    YEAR = { 2018 },
    MONTH = { October },
    PDF = { https://hal.archives-ouvertes.fr/hal-01726155/file/CROC_iros18.pdf },
    HAL_ID = { hal-01726155 },
    HAL_VERSION = { v2 },
    ABSTRACT = { We present a novel method for computing centroidal dynamic trajectories in multi-contact planning context. With dynamic motion it is necessary to respect kinematic and dynamic constraints during the contact planning step. Verifying the feasibility of a transition between contacts increase the success rate of the motion generation along the planned contacts. Our approach is based on a conservative but convex reformulation of the problem where we represent the center of mass trajectory as a Bezier curve, with control points constrained by the initial and final states and one free control point. Thanks to the convexity of this formulation, we can solve it efficiently with a Linear Program of low dimension. We use this LP as a feasibility criterion to test the contact transition candidates during multi-contact planning. By incorporating this criterion in an existing sampling-based contact planner, we are able to produce more robust contact sequences. We illustrate this application on various multi-contact scenarios. We also show that we can compute valuable initial guess, used to warm-start non-linear solvers for motion generation methods. This method could also be used for the 0 and 1-Step capturability problem. },
}