The International Journal of Robotics Study, Ahead of Publish.
This paper reports an unique outcome: with correct robotic versions based upon geometric auto mechanics, one can develop the kinodynamic movement preparation troubles for inflexible body systems as precise polynomial optimization troubles. Because of the nonlinear inflexible body characteristics, the movement preparation issue for inflexible body systems is nonconvex. Existing international optimization-based approaches do not parameterize 3D inflexible body movement successfully; hence, they do not scale well to long-horizon preparation troubles. We utilize Lie teams as the setup room and use the variational integrator to develop the forced inflexible body characteristics as square polynomials. After that, we utilize Lasserre’s pecking order of minute leisure to acquire the around the world ideal service by means of semidefinite shows. By leveraging the sparsity of the movement preparation issue, the recommended formula has straight intricacy relative to the preparation perspective. This paper shows that the recommended approach can offer around the world ideal options or certifications of infeasibility at the second-order leisure for 3D drone touchdown making use of complete characteristics and inverted kinematics for serial manipulators. Furthermore, we prolong the formulas to multi-body systems by means of the constricted variational integrators. The screening situations on cart-pole and drone with cable-suspended tons recommend that the recommended formulas can offer rank-one ideal options or nontrivial first assumptions. Ultimately, we suggest methods to quicken the calculation, consisting of an alternate solution making use of quaternion, which supplies empirically limited leisures for the drone touchdown issue at the first-order leisure.
发布者：Sangli Teng，转转请注明出处：https://robotalks.cn/convex-geometric-motion-planning-of-multi-body-systems-on-lie-groups-via-variational-integrators-and-sparse-moment-relaxation/
