Provably Stabilizing Controllers for Quadrupedal Robot Locomotion on Dynamic Rigid Platforms

DRS-LIP: Linear Inverted Pendulum Model for Legged Locomotion on Dynamic Rigid Surfaces

“…To highlight the usefulness of the analytical results, they have been used as a basis to synthesize a hierarchical planner that efficiently produces desire, physically feasible motions for quadrupedal robot walking on a DRS. The feasibility of the planned motion is validated by using our previous trajectory tracking controller [5] to command a quadrupedal robot to follow the planned motion during DRS walking. As discussed in Sec.V-B, both simulation and experimental results indicate the reasonable feasibility of the proposed planner under different gait parameters and surface motions (see Figs 7 and 8).…”

“…Legged robots have the potential to traverse various challenging surfaces, including stationary (uneven or discrete) surfaces [1]- [4] and dynamic rigid surfaces (i.e., rigid surfaces that move in the inertial frame) [5], [6]. Common real-world examples of the dynamic rigid surface (DRS) are ships, public transportation vehicles, trains, and elevators.…”

DRS-LIP: Linear Inverted Pendulum Model for Legged Locomotion on Dynamic Rigid Surfaces

“…This study aims to model and analyze the essential dynamic behaviors of a legged robot that walks on a DRS by building a reduced-order model and deriving its analytic approximate solution, so as to provide physical insights into the robot dynamics as well as to produce analytic results that can be used for efficient planning of legged locomotion. Yet, reduced-order modeling of DRS locomotion is fundamentally complex due to the nonlinear and high-dimensional nature of legged locomotion dynamics [7] and the timevarying movement of the surface-foot contact points [5].…”

Reinforcement Learning with Evolutionary Trajectory Generator: A General Approach for Quadrupedal Locomotion

“…The research and development of the quadrupedal robot have attracted much attention due to its potential to explore various terrains. Several approaches have achieved versatile locomotion skills in the quadrupedal robot, by designing the foot trajectories and the corresponding controller manually [5], [7], [9]. However, the performance of the designed controller relies on the expert knowledge of both the robotics system and the desired locomotion skill, and the development pipeline often involves tedious manual tuning [19], [24].…”