Co-design of a wheelchair-mounted robotic arm for functional autonomy: a Cybathlon case study in user-driven assistive robotics

Abstract Background Wheelchair-mounted robotic arms may support functional autonomy for individuals with upper-limb disabilities, but their long-term usability remains limited by the difficulty of matching high-dimensional robotic control to heterogeneous motor capabilities, daily routines, and task contexts. Participatory and user-centered design approaches are therefore relevant to assistive manipulation, provided that their design logic and limitations are reported transparently. Case presentation We report a longitudinal single-case co-design study of a wheelchair-mounted robotic arm developed with a 32-year-old man with C5–C6 spinal cord injury, who served both as pilot and collaborator during preparation for the 2024 Cybathlon Robotic Arm (ROB) competition. Starting from an off-the-shelf robotic manipulator (Kinova Gen3), the system evolved through one year of weekly task trials, user feedback, technical logging, video review, and team debriefings. Methods The study is presented as a reflexive system-development case study rather than as a controlled efficacy trial or formal qualitative analysis. Mechanical integration, interface layout, and control policies were progressively adapted through a documented decision trail linking observed interaction breakdowns to design responses. Two control adaptations are analyzed post hoc: (1) a hand-motion interface for translational control, and (2) an adjustable orientation coupling, referred to as “snake mode”, designed to reduce explicit manual rotational commands during selected tasks. Results In selected illustrative trials, hand-motion control showed movement patterns consistent with more continuous translational control than tablet joystick input, whereas snake mode reduced explicit manual rotational commands during an eating task. During the Cybathlon competition, the final system achieved competitive task performance and low failure rates while preserving continuous pilot control. These results support the functional maturity of the co-designed system in a standardized activity-based benchmark, but do not establish general usability or long-term adoption. Conclusion This single-case study suggests that sustained co-design can support robust assistive manipulation when control allocation is explicitly matched to user capabilities and when assistance remains transparent to the operator. Broader multi-user studies with standardized usability, workload, and longitudinal home-use measures are required before general claims can be made.