Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An adaptive hierarchical control framework for quadrupedal robots in planetary exploration

Franek Stark, Rohit Kumar, Shubham Vyas, Hannah Isermann, Jonas Haack, Mihaela Popescu, Jakob Middelberg, Dennis Mronga, Frank Kirchner

TL;DR

This work tackles robust quadrupedal locomotion in unknown planetary environments by integrating adaptive gait sequencing, online mass/COM adaptation, and whole-body control within a modular ROS 2 framework. The approach combines bio-inspired gait adaptation, online model parameter updates via Kalman filtering, and MPC/WBC-based planning to handle uncertainties in robot dynamics and terrain. Key contributions include a runtime-reconfigurable software architecture, an adaptive gait sequencer, online mass/COM estimation, and a contact-aware state estimator, all validated on two robots and in a volcanic analog field test exceeding 700 m. The results demonstrate reduced falls, improved energy efficiency, and more accurate tracking under payload changes, highlighting the framework’s potential for autonomous, long-duration planetary exploration with legged robots.

Abstract

Planetary exploration missions require robots capable of navigating extreme and unknown environments. While wheeled rovers have dominated past missions, their mobility is limited to traversable surfaces. Legged robots, especially quadrupeds, can overcome these limitations by handling uneven, obstacle-rich, and deformable terrains. However, deploying such robots in unknown conditions is challenging due to the need for environment-specific control, which is infeasible when terrain and robot parameters are uncertain. This work presents a modular control framework that combines model-based dynamic control with online model adaptation and adaptive footstep planning to address uncertainties in both robot and terrain properties. The framework includes state estimation for quadrupeds with and without contact sensing, supports runtime reconfiguration, and is integrated into ROS 2 with open-source availability. Its performance was validated on two quadruped platforms, multiple hardware architectures, and in a volcano field test, where the robot walked over 700 m.

An adaptive hierarchical control framework for quadrupedal robots in planetary exploration

TL;DR

This work tackles robust quadrupedal locomotion in unknown planetary environments by integrating adaptive gait sequencing, online mass/COM adaptation, and whole-body control within a modular ROS 2 framework. The approach combines bio-inspired gait adaptation, online model parameter updates via Kalman filtering, and MPC/WBC-based planning to handle uncertainties in robot dynamics and terrain. Key contributions include a runtime-reconfigurable software architecture, an adaptive gait sequencer, online mass/COM estimation, and a contact-aware state estimator, all validated on two robots and in a volcanic analog field test exceeding 700 m. The results demonstrate reduced falls, improved energy efficiency, and more accurate tracking under payload changes, highlighting the framework’s potential for autonomous, long-duration planetary exploration with legged robots.

Abstract

Planetary exploration missions require robots capable of navigating extreme and unknown environments. While wheeled rovers have dominated past missions, their mobility is limited to traversable surfaces. Legged robots, especially quadrupeds, can overcome these limitations by handling uneven, obstacle-rich, and deformable terrains. However, deploying such robots in unknown conditions is challenging due to the need for environment-specific control, which is infeasible when terrain and robot parameters are uncertain. This work presents a modular control framework that combines model-based dynamic control with online model adaptation and adaptive footstep planning to address uncertainties in both robot and terrain properties. The framework includes state estimation for quadrupeds with and without contact sensing, supports runtime reconfiguration, and is integrated into ROS 2 with open-source availability. Its performance was validated on two quadruped platforms, multiple hardware architectures, and in a volcano field test, where the robot walked over 700 m.
Paper Structure (11 sections, 16 equations, 4 figures, 1 table)

This paper contains 11 sections, 16 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Adaptive walking controller
  3. Adaptive Gait Sequencer
  4. Whole Body Control (WBC)
  5. Kalman Filtering (KF) for Model Adaptations
  6. State Estimation with Floating Base and Contacts in Planetary Scenarios
  7. Results
  8. Computational evaluation
  9. Adaptive Gait Sequencer
  10. Model Adaptation
  11. Discussion and Conclusion

Figures (4)

  • Figure 1: The control framework was successfully tested on both a Unitree Go2 robot and the custom build quadrupedal robot B12 during a field test to Vulcano Island in Italy.
  • Figure 2: Block diagram showing the architecture of the control framework
  • Figure 3: Cost of transport over velocity of the two gait sequencers during trotting on the field trip. The plot shows the combined data of all experiments. The $r^2$ values for each fit are plotted onto the graph.
  • Figure 4: a) Estimated total mass with ground truth (top) and estimated COM (bottom) during dynamic payload switching experiment. The original robot weight is 16.21 and the COM offsets are $c_x =$ 8.8, and $c_y =$ 0.0. The time intervals where a specific amount of payload (displayed above the upper graph) is attached to the robot are highlighted in grey. b) Height and orientation error of nominal controller (left) and adaptive controller with KF (right) over dynamic weight switching experiment.