Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cybathlon -- Legged Mobile Assistance for Quadriplegics

Carmen Scheidemann, Andrei Cramariuc, Marco Hutter

TL;DR

This work presents an independent quadrupedal assistive platform combining an ANYmal D base with a DynaArm and a mouth-controlled QuadStick, augmented by a voice-command interface, to perform everyday tasks for quadriplegics. It employs both a model-based MPC end-effector controller for precise manipulation and a learned RL-based locomotion policy for base movement, enabling flexible control while maintaining safety via autonomous face-touching automation. In Cybathlon 2024, the system achieved a full score with best times around $6$ minutes, demonstrating strong performance and generalization to unseen tasks such as scarf hanging. The results emphasize the platform’s potential to provide independence beyond wheelchair mobility, with broader implications for future autonomous, certifiable assistive robots.

Abstract

Assistance robots are the future for people who need daily care due to limited mobility or being wheelchair-bound. Current solutions of attaching robotic arms to motorized wheelchairs only provide limited additional mobility at the cost of increased size. We present a mouth joystick control interface, augmented with voice commands, for an independent quadrupedal assistance robot with an arm. We validate and showcase our system in the Cybathlon Challenges February 2024 Assistance Robot Race, where we solve four everyday tasks in record time, winning first place. Our system remains generic and sets the basis for a platform that could help and provide independence in the everyday lives of people in wheelchairs.

Cybathlon -- Legged Mobile Assistance for Quadriplegics

TL;DR

This work presents an independent quadrupedal assistive platform combining an ANYmal D base with a DynaArm and a mouth-controlled QuadStick, augmented by a voice-command interface, to perform everyday tasks for quadriplegics. It employs both a model-based MPC end-effector controller for precise manipulation and a learned RL-based locomotion policy for base movement, enabling flexible control while maintaining safety via autonomous face-touching automation. In Cybathlon 2024, the system achieved a full score with best times around minutes, demonstrating strong performance and generalization to unseen tasks such as scarf hanging. The results emphasize the platform’s potential to provide independence beyond wheelchair mobility, with broader implications for future autonomous, certifiable assistive robots.

Abstract

Assistance robots are the future for people who need daily care due to limited mobility or being wheelchair-bound. Current solutions of attaching robotic arms to motorized wheelchairs only provide limited additional mobility at the cost of increased size. We present a mouth joystick control interface, augmented with voice commands, for an independent quadrupedal assistance robot with an arm. We validate and showcase our system in the Cybathlon Challenges February 2024 Assistance Robot Race, where we solve four everyday tasks in record time, winning first place. Our system remains generic and sets the basis for a platform that could help and provide independence in the everyday lives of people in wheelchairs.
Paper Structure (12 sections, 4 figures, 1 table)

This paper contains 12 sections, 4 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. SYSTEM
  3. ANYmal with an Arm
  4. Moving the System
  5. Interfacing with the Pilot
  6. Quadstick Interface
  7. Mode-Independent Inputs
  8. Mode-Specific Inputs
  9. Autonomous Face Touching
  10. Voice Command Interface
  11. RESULTS
  12. CONCLUSIONS

Figures (4)

  • Figure 1: A overview of our proposed system in operation during the Cybathlon Challenges February 2024. The main components of the system are an ANYmal quadrupedal base and a DynaArm robotic arm. Pictured is the pilot operating the robot to open a mailbox, which is the first task of the racetrack.
  • Figure 2: The initial configurations of the two control modes of the Ocs2 controller. The left Figure shows that of EE Control Mode Front, the right that of EE Control Mode Top.
  • Figure 3: Quadstick Layout and Shared Input Assignments
  • Figure 4: Visualizations of the four tasks featured in the Cybathlon 2024 February Challenge Robotic Assistance Race. From left to right, the tasks are: (i) open a mailbox, pick the package inside, and place it on the table; (ii) pick up the toothbrush, bring it to the pilot's mouth, and place it back in the cup; (iii) pick up the scarf and hang it on the clothes rack; and (iv) open the dishwasher, retrieve a plate from inside, and place it on top.