Online DCM Trajectory Generation for Push Recovery of Torque-Controlled Humanoid Robots

Abstract

We present a computationally efficient method for online planning of bipedal walking trajectories with push recovery. In particular, the proposed methodology fits control architectures where the Divergent-Component-of-Motion (DCM) is planned beforehand, and adds a step adapter to adjust the planned trajectories and achieve push recovery. Assuming that the robot is in a single support state, the step adapter generates new positions and timings for the next step. The step adapter is active in single support phases only, but the proposed torque-control architecture considers double support phases too. The key idea for the design of the step adapter is to impose both initial and final DCM step values using an exponential interpolation of the time varying ZMP trajectory. This allows us to cast the push recovery problem as a Quadratic Programming (QP) one, and to solve it online with stateof-the-art optimisers. The overall approach is validated with simulations of the torque-controlled 33 kg humanoid robot iCub. Results show that the proposed strategy prevents the humanoid robot from falling while walking at 0.28 m/s and pushed with external forces up to 150 Newton for 0.05 seconds.

Online DCM Trajectory Adaptation for Push and Stumble Recovery during Humanoid Locomotion

“…where we used the identities α DS (T DS ) = 0 and γ DS (T DS ) = 1. As Ṽi+1 is the same as Ṽankle during the following single support phase, the result in ( 21) is identical to the one obtained by evaluating (15) for t = 0. On the other hand, Ξankle is discontinuous at the end of the single support phase and the start of the following double support phase (see Fig.…”

“…A footstep adaptation algorithm based on DCM dynamics and using an exponential interpolation of the ZMP trajectory was proposed in [15], but the step adaptation was active only during the single support phase. Jeong et al [16] presented a robust walking controller based on the DCM dynamics using an online optimization solver to combine the ankle, hip, and stepping strategies.…”

Simulation of Disturbance Recovery Based on MPC and Whole-Body Dynamics Control of Biped Walking

“…Throughout the walking process, the ZMP is located in the stable support area, achieving stable control. The ability to reject 400 N and last 0.1 s external force in this paper is better than literature [20], which is 350 N and 0.05 s, respectively, and it also better than literature [21], which is 6.0 Nm. This simulation proves that when the model is disturbed by external forces, it can quickly recover the original stable walking state and achieve the desired goal.…”

A Benchmarking of DCM-Based Architectures for Position, Velocity and Torque-Controlled Humanoid Robots

“…Another interesting future work is the implementation of a footstep adjustment algorithm. 43,44,45 This will increase the overall robustness in case of large disturbances acting on the robot.…”