MYOLINK esports: Exploring EMG-based Control Interface through Muscle Activation and Inhibition to Enable Common Gameplay Mechanics among Players with and without Physical Disabilities
Masato Shindo, Shiina Takano, Shuto Sako, Akihiro Miyata, Ryosuke Aoki · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3791917
Summary
Shindo, Takano, Sako, Miyata, and Aoki (NTT and Nihon University) propose MYOLINK esports, a paradigm for inclusive competitive gaming that shifts game control from kinematic inputs (body position and posture) to kinetic inputs (muscle force). Four surface-EMG sensors are placed on muscles of the player's choosing; gameplay commands (right, left, forward, backward) are triggered by voluntary activation and inhibition of each muscle, with per-user thresholds set by calibration. The authors argue that by removing the requirement for specific body movements — and instead asking players to coordinate activation/inhibition of whichever muscles they can voluntarily control — competitive play becomes accessible to players with a wide range of physical disabilities while remaining the same interface for non-disabled players. Two user studies evaluate the system using Rocket League (a vehicular soccer esports title): User Study 1 with 20 able-bodied participants examined whether movement magnitude (wrist/elbow angle) affects EMG control accuracy; User Study 2 with 8 disabled participants (cervical spinal cord injury, spinal muscular atrophy, upper-limb amputation) evaluated whether EMG-IF control accuracy differs between disabled and able-bodied players. Two additional case studies document hands-on exhibition play at mixed-ability events and a custom cooperative "Jump Game," and a drone-race game (MYOLINK Drone Race) is introduced as an EMG-native esports title.
Key findings
Command-input accuracy in EMG-IF was not significantly correlated with the range of wrist movement, and only weakly (inversely) with elbow movement — suggesting that the interface is substantially decoupled from body-movement magnitude, which is the core accessibility claim. Across User Studies 1 and 2, the only statistically significant between-group difference in temporal steering accuracy was in the low-tempo steering-only condition (p = 0.038, r = 0.33); the other five conditions showed no significant difference between disabled and able-bodied players. Ball-approach accuracy in the free-game task showed no significant group difference but a large effect size (Cohen's d = 0.80) favouring disabled participants, which the authors interpret as possibly reflecting more selective muscle recruitment developed through daily SCI adaptation. Where EMG-IF struggled was with disability-related involuntary muscle activity: strong spasticity (P22) and unintended co-contraction (P28) repeatedly triggered wrong commands. Practical contributions include: calibration-based personalisation of sensor placement and thresholds (the authors compare it to a motorsport "pit stop"), acknowledgement that current low-cost HCI EMG sensors lack the fidelity of medical-grade kit (Delsys Trigno) used in the study, and a pathway for mixing conventional and EMG controllers in the same match.
Relevance
For accessibility practitioners working on gaming, input, or assistive technology, this paper offers a well-designed empirical case for "kinetic" inputs as a common interface across disabled and non-disabled players, avoiding the separatism of dedicated adaptive controllers while still accommodating severe motor impairment. The candid discussion of spasticity, co-contraction, fatigue, and signal-to-noise limits is especially useful — rare in accessible-gaming papers, which often overclaim universality. Design implications beyond esports include EMG as complementary input for users of smart speakers, VR, and AAC, and the warning that affordable consumer EMG devices may mislead users about their own biomechanics if not paired with expert guidance. Limitations: small disabled sample (N=8); no participants with severe spasticity outside P22; no participants with very low EMG amplitude (ALS, locked-in syndrome) where the approach may not work at all; reliance on expensive medical-grade sensors; and a single esports title (Rocket League) limits generalisability to fighting, rhythm, or FPS genres.
Tags: esports · inclusive esports · electromyography · EMG · game accessibility · physical disabilities · spinal cord injury · spinal muscular atrophy · adaptive gaming · muscle-based input