Overview

This was a team project in which we upgraded an existing Ackerman vehicle by turning it into an Automated Guided Vehicle. This was achieved by using ROS as framework on a NVIDIA Jetson Xavier computer and with the implementation of navigation concepts such as mapping, path planning and motion control. As a result, we managed to navigate the AGV inside a closed environment by setting a goal position and avoiding collisions.

My main role was as software developer, although I also contributed to hardware adaptations whenever we faced challenges translating our software output to the vehicle.

Tools

Hardware

  • RPLIDAR A3: laser sensor for mapping and object detection.
  • NVIDIA Jetson Xavier: main computer running all the software.
  • Arduino Mega 2560: microcontroller board.
  • DC motors: movement of the AGV steering and acceleration.
  • DC-DC converter modules: DC motor and NVIDIA Jetson Xavier power supply.
  • 5V Step-Down converter: power supply for acceleration and steering H-bridges, relay module, and RC receiver.
  • RC controller: controlling AGV in manual mode.

Software

  • Python: programming language for communication between TEB local planner output and Arduino IDE.
  • Arduino: programming language for embedded software development.
  • Arduino IDE: software environment for writing and running Arduino code.
  • ROS: framework for robot communication.
  • Hector SLAM: mapping of CIMA space.
  • GIMP: map refinement and editing.
  • RViz: visualization of map and robot navigation.
  • TEB local planner: navigation algorithm.
  • SolidWorks: CAD of RPLIDAR A3 and NVIDIA Jetson Xavier mounting bases.
  • Alltrax 39 Controller Pro: traction motor controller.

Key Improvements

  • Damaged steering motor replacement.
  • Electronic circuit organization.
  • Attachment of the NVIDIA Jetson Xavier and RPLIDAR A3 to the AGV using custom laser-cut bases.
  • Mapped an engineering laboratory called CIMA.
  • Added the navigation system by using a RPLIDAR A3 and Hector SLAM.
  • Added path planning with TEB local planner, with the desired linear and angular velocities as output.
  • Added a toggle to switch between autonomous and manual mode with an RC controller in case human intervention was needed.

Video