Balance control based on Capture Point error compensation for biped walking on uneven terrain

Abstract

This paper tries to improve a balance control based on the Capture Point (CP) control. First the characteristics of the conventional balance controller are shown to be essentially the same as the CP controller. Then we analyze the transfer function of the balance controller. We introduce a new state variable with the CP integration to the CP and the ZMP (Zero-Moment Point) in order to trim a long term offset of the CP and the ZMP. Verification of the proposed balance controller is conducted through both simulation and experiments with a humanoid robot HRP-2[11].

Walking on gravel with soft soles using linear inverted pendulum tracking and reaction force distribution

“…Our ZMP-COM control layer is based on [11], [12], [13]. We define a feedback controller on the state…”

“…We then proceed by pole placement in order to obtain the best COM-ZMP regulator [11], which is equivalent to a capture-point tracking controller [13].…”

“…To account for joint flexibilities and sole compliance, we assume that the real ZMP of the robot (P ZMP ) lags behind the control ZMP (P c ZMP ). As in [12], [13], we model this delay as a low-pass filter whose transfer function is:…”

Robot-human balance state transfer during full-body humanoid teleoperation using Divergent Component of Motion dynamics

“…The work in [7] and [10] use the DCM of a model with variable natural frequency in order to regulate the robot linear and angular momentum for stable standing and walking under external disturbances. Furthermore, the work by [11] is inspired on the optimal regulator in order to design a CoM/CoP controller inspired by the DCM. These authors explore the concept of the stabilization of the equilibrium state of the CoM in order to improve disturbance robustness.…”

Robust Phase-Space Planning for Agile Legged Locomotion over Various Terrain Topologies

“…Sharing similar core ideas, the divergent component of motion [15] and the extrapolated center-of-mass [16] were independently proposed. Extensions of the Capture Point method [17,18], allow locomotion over rough terrains. Recently, the work in [19] generalizes the Capture Point method by proposing a "Nonlinear Inverted Pendulum" model, but it is limited to the two-dimensional case, and angular momentum control is ignored.…”