Mojito introduces a novel approach to human motion analysis using lightweight, privacy-conscious Inertial Measurement Units (IMUs) enhanced by large language models.

• Overcomes traditional challenges in IMU data processing, particularly jitter reduction in wireless streaming
• Provides a privacy-conscious alternative to vision-based motion tracking systems
• Enables accurate capture of dynamic forces and torques in 3D human motion
• Leverages LLMs to interpret and contextualize complex motion data

Medical Impact: Mojito's wearable IMU approach offers significant potential for monitoring physical activity, supporting rehabilitation programs, and analyzing movement disorders with greater privacy and accessibility than camera-based solutions.

Mojito: LLM-Aided Motion Instructor with Jitter-Reduced Inertial Tokens