Mojito introduces a novel approach that integrates inertial measurement units (IMUs) with large language models to capture and interpret human bodily movements with reduced jitter.

• Provides privacy-conscious motion sensing as an alternative to vision-based systems
• Processes streaming IMU data to capture dynamic forces and torques inherent in 3D motion
• Addresses technical challenges of wireless signals, including jitter reduction
• Offers lightweight, wearable monitoring solution

Medical Impact: This technology enables improved monitoring of physical activity, supports rehabilitation processes, and enhances analysis of movement disorders—all while maintaining patient privacy and comfort through non-intrusive wearable technology.

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