We developed an automated camera system to address the challenge of filming MMA grappling matches, where athletes move rapidly and remain in close contact through rolls, turns, and frequent position changes. Traditional filming requires multiple camera operators to capture the action without missing critical moments, creating a significant operational burden.
Our team designed a three-camera system mounted on motorized tripods controlled by a central processing unit. Each camera operates on battery power, eliminating cable requirements, and consists of a camera module, microcontroller, and automated pan-tilt mount. We implemented AI-based athlete detection and tracking algorithms that directs the cameras to follow competitors autonomously. The central device aggregates video feeds from all three cameras and selects the optimal view for recording and display in real time.
We tested the system during live training sessions with the FSU Brazilian Jiu-Jitsu club. The cameras successfully tracked rapid movements and ground fighting while maintaining stable, centered framing of the athletes. The system avoided abrupt camera movements and frame instability, producing clear video that captured important moments without gaps in coverage.
Our results demonstrate that AI-driven camera systems can effectively automate combat sports filming, reducing the need for multiple human operators while maintaining video quality. The system keeps athletes properly framed and visible throughout matches, producing footage suitable for viewing and analysis.
