Overview
RecycloBot was developed during a three-month engineering challenge. The objective was to build a robot capable of collecting empty cans from a table and placing them in a designated area.
Our team designed and integrated the complete system, including the 3D-printed structure, custom PCB, sensors, and FreeRTOS-based firmware. For the high-level manipulation task, we trained a neural network with reinforcement learning in Unity ML-Agents and deployed it on the embedded platform using STM32Cube.AI.
Training
Gallery
