Design and build of an affordable DIY system for field motion capture: a volleyball spike field test

Abstract

Motion capture is widely used to refine sports technique and to guide recovery from injury, but laboratory systems are bulky and too expensive for most hobby athletes, schools and clubs. We designed an open-source kit that pairs a wrist-worn inertial sensor, built from an Arduino Nano 33 BLE, with a smartphone camera processed by MediaPipe Pose, and tested it on 139 volleyball spikes from 12 players outdoors. The wrist unit costs 40 USD in consumer parts and can be assembled with basic tools. The camera channel tracked the hitter in every analysed spike and met its frame-timing thresholds in 95 percent of them. The wrist sensor delivered a usable stream in 42 of the 135 analysed spikes, at 53 to 92 samples per second, and resolved swings near 2000 degrees per second; most of the losses arose in the Bluetooth streaming chain rather than in the sensor itself. The camera-derived elbow and wrist features took values in a plausible range across the dataset. These results show that the system can record a complex whole-body movement outdoors, that the wrist sensor delivers high-rate data when its stream arrives intact, and that the camera channel alone yields arm features in a plausible range. They also point to local data logging as the change most likely to make the sensor channel as dependable as the camera. We publish a blueprint for the build together with the data and code.