<![CDATA[Advances in computer vision technology enable human-computer interaction without relying on conventional input devices such as keyboards or mice. One prominent technology being widely developed is MediaPipe-based hand tracking, which is capable of mapping hand positions in real-time using an RGB camera. This research implements a hand tracking system using Python and MediaPipe as an alternative control method in a delivery simulation game titled Tofu Deliver. The game was developed using the Unity engine and features a core gameplay concept centered around the production and distribution of tofu to customers. The system architecture incorporates two control modes—keyboard and hand tracking—integrated via TCP socket communication between Unity and Python. Evaluation was conducted to test the accuracy of gesture detection, system responsiveness, and user experience aspects. The implementation results indicate that the hand tracking system successfully drives the vehicle in real-time with optimal response levels, thereby offering a more natural interaction preference compared to conventional control schemes.]]>
Copyrights © 2026
