A Proprioceptive, Force-Controlled, Non-Anthropomorphic Biped for Dynamic Locomotion

Abstract

This paper addresses the challenge of terrainadaptive dynamic locomotion in humanoid robots, a problem traditionally tackled by optimization-based methods or reinforcement learning (RL). Optimization-based methods, such as model-predictive control, excel in finding optimal reaction forces and achieving agile locomotion, especially in quadruped, but struggle with the nonlinear hybrid dynamics of legged systems and the real-time computation of step location, timing, and reaction forces. Conversely, RL-based methods show promise in navigating dynamic and rough terrains but are limited by their extensive data requirements. We introduce a novel locomotion architecture that integrates a neural network policy, trained through RL in simplified environments, with a state-of-the-art motion controller combining model-predictive control (MPC) and whole-body impulse control (WBIC). The policy efficiently learns high-level locomotion strategies, such as gait selection and step positioning, without the need for full dynamics simulations. This control architecture enables humanoid robots to dynamically navigate discrete terrains, making strategic locomotion decisions (e.g., walking, jumping, and leaping) based on ground height maps. Our results demonstrate that this integrated control architecture achieves dynamic locomotion with significantly fewer training samples than conventional RLbased methods and can be transferred to different humanoid platforms without additional training. The control architecture has been extensively tested in dynamic simulations, accomplishing terrain height-based dynamic locomotion for three different robots.

Sideways crab-walking is faster and more efficient than forward walking for a hexapod robot

“…Sideways walking has been studied in a relatively smaller number of robots. One important exception in bipeds [49] shows that rotating the hips and knees to operate in the sagittal plane with a nonanthropomorphic gait is akin to walking sideways and reduces roll oscillations [50]. In this design, adding rotation at the hip is best for steering, resulting in a biped leg design that is similar to a crab leg [51].…”

Leg Trajectory Planning for Quadruped Robots with High-Speed Trot Gait

“…The proprioceptive actuator paradigm achieves a combination of high-bandwidth force control, high torque density, as well as impact mitigation [9]. For example, NABi-V2 utilizes six-back drivable high-power density electromagnetic actuators with a low gear ratio single-phase planetary gearbox to realize proprioceptive, force-controlled dynamic locomotion [12]. Similarly, a single-stage planetary gear reduction (main material is titanium alloy, and weight is 0.113 kg) with low gear ratio (6:1) combined with high torque density frameless motors was applied during our leg design.…”

Gait Event Detection with Proprioceptive Force Sensing in a Powered Knee-Ankle Prosthesis: Validation over Walking Speeds and Slopes

“…These so-called proprioceptive actuators can exploit an algebraic relationship between actuator torque and ground contact force, thus making exteroceptive sensors unnecessary [14]. This type of actuator has been utilized in the literature for the application of proprioceptive force control to achieve quadrupedal bounding [14] and bipedal pronking motions [17]. Recently, quasi-direct drive actuators have been implemented in powered prosthetic legs [7], [18], but the concept of force proprioception has not yet been explored in this context.…”