On the Hardware Feasibility of Nonlinear Trajectory Optimization for Legged Locomotion based on a Simplified Dynamics

Abstract

We propose two feasibility constraints to be included in a Single Rigid Body Dynamics (SRBD)-based trajectory optimizer in order to obtain robust motions in challenging terrain. The former finds an approximate relationship between joint torque limits and admissible contact forces without requiring the joints' configuration. The latter proposes a model of the leg to guarantee the avoidance of the collision with the environment. Such constraints have been included in a nonlinear non-convex optimization problem. We validate the feasibility of the trajectories both in simulation and on the Hydraulically actuated Quadruped (HyQ) robot, including experiments with non flat terrain.

Tracking Locomotion using Reinforcement Learning

“…The power of this strategy is harnessed through realistic on-board sensors and simulations of the hydraulically actuated four-legged robot HyQ Real that traverses rugged terrain under computational conditions. 4) Angelo Bratta, Romeo Orsolino, Michele Focchi , Victor Barasoul et al [4] The proposed leg-locomotion nonlinear trajectory optimization based on simplified dynamics helps to achieve robust locomotion in difficult terrain constraints and focuses on incorporating joint torque limiting constraints, these limitations. Approximate how it is converted to permissible contact force as it is mapped to the leg.…”

“…Searching for sample-efficient methods (that is, methods that require little data) generally requires the development of increasingly complex methods. This increased complexity poses a reproducibility crisis [4]. Recent studies have shown that many RL methods are not robust to hyper parameter changes, random starting values, or different implementations of the same algorithm [12,13].…”