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

International Journal of Humanoid Robotics (IJHR) 2019

Giulio Romualdi, Stefano Dafarra, Yue Hu, Prashanth Ramadoss, Francisco Javier Andrade Chavez, Silvio Traversaro, Daniele Pucci

Overview

The Divergent Component of Motion (DCM) is a cornerstone of many bipedal walking architectures. This paper provides a systematic benchmark of DCM-based control architectures, evaluating how they behave when the underlying robot is driven in position, velocity, or torque control mode.

Method

  • Unified architecture. A layered DCM-based locomotion architecture is implemented so the same high-level planner can drive different low-level control modes.
  • Comparative study. Position-, velocity-, and torque-controlled variants are evaluated under consistent conditions to expose their trade-offs.

Results

The benchmark clarifies the strengths and limitations of each control mode for DCM-based walking on humanoid robots. The full implementation is released as the open-source walking-controllers project.

← Back to all publications