T.A.C.T. – Tactical Augmented Coaching Translator

T.A.C.T, short for Tactical Augmented Coaching Translator, is an augmented reality basketball training system designed to bridge the gap between how tactics are explained and how they are executed on court. The project targets basketball academies and young trainees who often struggle to understand complex tactic-board drawings and convert them into real movement.

The system transforms a coach’s 2D tactical explanation into an interactive AR training scenario. The pipeline begins when the coach uploads a video explanation on a court board. AI-based analysis then extracts the players, positions, timing, and basketball actions into a structured JSON format. This output is executed in Unity, where each action is mapped to animation clips and synchronized with virtual players. The system also supports persistent AR court placement using spatial anchoring and manual calibration, allowing the training scene to stay aligned with the real environment.

A key contribution of T.A.C.T is closing the training loop through real-time feedback. A rover-based computer vision system, using a Raspberry Pi, camera, and WebSocket communication, tracks the player and supports dynamic coaching flow. The system verifies actions using posture analysis and provides feedback on positioning, passing, shooting, and receiving. In addition, the free-throw module applies pose estimation to evaluate shooting form and generate performance feedback.

Overall, T.A.C.T combines robotics, augmented reality, artificial intelligence, and computer vision to create a more interactive, measurable, and engaging basketball coaching experience.

Project Details

[photo]