This thesis presents the design and implementation of a system for wireless control of a car using hand gesture recognition based on machine learning. The system consists of two ESP32-based devices: a glove with a motion sensor and a modified car from the ELEGOO Smart Robot Car V4.0 kit. The device worn on the hand captures data about hand movements and wirelessly transmits it to the car. The car can operate in two modes: a normal mode for basic direction control in which the car moves, and an advanced mode that performs more complex movements based on predefined gestures. A machine learning model was developed using TensorFlow to classify the gestures, then converted to TensorFlow Lite Micro and deployed on the ESP32. Based on the motion sensor data, the model recognizes gestures such as circles, spinning, or zig-zag movements and controls the car accordingly. The thesis demonstrates the successful integration of embedded systems, wireless communication, sensor data processing, and machine learning on resource-constrained devices.
|
