Project Overview

Developed the control and navigation system for ADIUTOR, an autonomous differential-drive mobile robot designed for hospital logistics. The project combines an STM32-based embedded motor-control architecture with a ROS/Gazebo autonomous navigation system. Using RPLIDAR A2 data and the Input Space Sampling (ISS) method, the robot detects obstacles in real time, evaluates collision-free motion candidates, and selects the trajectory that best approaches the target without requiring prior global path planning.

Technical Highlights

Developed a reactive obstacle-avoidance algorithm based on Input Space Sampling (ISS).

• Processed 360° LiDAR data to classify front, left, and right regions as free or occupied.

• Used a differential-drive kinematic model, wheel odometry, and Euclidean-distance optimization to predict and select the next robot position.

• Implemented and tested the navigation system using ROS, Gazebo, Python, and RQT Graph.

• Validated navigation in indoor environments containing walls, doors, people, and multiple obstacles.

• Designed the embedded control architecture around an STM32F407VET6 ARM Cortex-M4 microcontroller.

• Implemented PWM motor-speed control, PID regulation, encoder acquisition through external interrupts, direction reversal, and braking control.

• Designed the interface for two 36 V brushless DC motors, Kelly KBS motor controllers, Omron E6B2 incremental encoders, and Hall-effect sensors.

• Integrated galvanic isolation, voltage regulation from 36 V to 5 V and 3.3 V, overvoltage protection, fuse protection, and an emergency-stop circuit.

• Configured peripherals using STM32CubeMX and developed the embedded firmware in STM32CubeIDE.

Key Result

The simulated robot successfully detected and avoided multiple obstacles in real time, selected collision-free trajectories, and reached a predefined target while maintaining a safety distance.