Idea
An app-based arm pose estimation platform using smartphone and smartwatch sensors for fitness, rehabilitation, and AR developers.
Research Paper
Core Innovation
This paper introduces SmartPoser, which uniquely combines UWB distance measurements with IMU data from common consumer devices to estimate arm pose. Unlike prior methods relying on cameras or multiple specialized sensors, it achieves accurate wrist and elbow position estimates without user training. This enables practical, low-cost arm pose tracking using only a smartphone and smartwatch.
Market Size (TAM)
$2–10B TAM, $1–2B SAM; assumption: growing demand for wearable motion tracking in fitness, healthcare, and AR/VR sectors.
Potential Customers & Pain Points
- Fitness app developers needing accurate motion tracking
- Physical therapists requiring affordable arm movement monitoring
- AR/VR companies seeking low-cost pose estimation
- Wearable device makers wanting enhanced sensor fusion
- Sports coaches needing real-time arm position feedback
Business Model
Licensing SDK/API to app developers and wearable manufacturers; subscription for advanced analytics and cloud processing; partnerships with healthcare and fitness platforms
Competitive Landscape
- Xsens
- Notch
- Perception Neuron
Implementation Challenges
- Limited accuracy compared to multi-camera systems
- Dependence on UWB hardware availability
- User variability in device placement affecting results
Validation Strategy
- Develop prototype app integrating UWB and IMU data
- Conduct user studies comparing against ground truth motion capture
- Pilot deployments with fitness and rehabilitation partners
Research Paper Overview
SmartPoser: Arm Pose Estimation with a Smartphone and Smartwatch Using UWB and IMU Data
Summary
This paper presents a method to estimate arm pose using only an off-the-shelf smartphone and smartwatch by combining ultra-wideband (UWB) distance measurements with inertial measurement unit (IMU) data. This approach overcomes limitations of prior systems that require cameras or multiple sensors by providing accurate wrist and elbow joint position estimates with a median error of 11.0 cm without user training.