MyoRun – Endless Runner Game controlled with Muscles

 

Worried about touching phones, keyboards, mouses in this time of COVID? Just get a couple electrode pads on your arms, and start playing games.

The project is a computer game controlled in Real Time by output from an Electromyography (EMG) system worn by the user. Arm movements/muscle flexes will be used to control movement in the game. With a combination of Digital Signal Processing (DSP) and Machine Learning (ML) applied to training data recorded using our EMG system, these signals are used as inputs into the game. The game is highly interactive and responsive so that the player will move accordingly when the user indicates their intended action with their arm movements. Our objective for this project is to extract meaning from EMG signals real-time and control the movement of a character in a game environment using those signals.

Design Goals:

• Finding muscle location for electrode placement
• Optimize number of muscle sensors to produce 4 outputs for game control
• DSP + ML Techniques to output at least 80% classification accuracy
• Designing game with less than 250 ms latency
• Overall End to End delay with less than 500 ms latency

Results

• 5 Locations on Arms including Ground
• Flexion, Extension, Pronation, Supination
• Support Vector Machine produces 86% accuracy
• Game Latency of 16.637ms
• Total Latency of 397.4314ms