Robot Pendeteksi Korosi Bagian Bawah Kendaraan Transportasi Publik berbasis Machine Learning dan IoT (2024 GEMASTIK Penyisihan)

This research project focuses on the development and testing of an IoT-based corrosion detection robot. Using YOLO (You Only Look Once) algorithm with OpenCV for high-accuracy corrosion detection on vehicle undersides, the robot processes images collected via a phone camera. The system includes data labeling, YOLO model training, and integration with hardware. Powered by an STM32 microcontroller and mini PC, it processes sensor data and machine learning algorithms. The robot’s movement is controlled via joystick using Bluetooth communication.

This prototype was developed for GEMASTIK 2024 in the IoT Development category. With this product, we reached the top 10 in the internal selection for joining national qualification round.

On this project, i have done mostly in kinematics and communication of the robot using ROS 1 (noetic) and STM32 as microcontroller. 

- For kinematics and movement, i implemented calculation for 3 wheels omni and PID control. 

- For communication, i used USB CDC FS for STM32 and serial library for ROS1.

- The robot's movement is controlled using a PS4 joystick, connected via Bluetooth to the ROS.