TerrainPilot Buggy

Tech stack

• Python
• C++
• ROS2
• Jetson Orin Nano
• Pandar40P LiDAR
• IMU
• GPS
• SLAM
• Sensor Fusion
• CAN Bus
• Linux
• OpenCV
• NumPy
• Motor Control
• 48V Battery System

Overview

A self driving off road vehicle platform designed to navigate unstructured terrain using LiDAR, GPS, IMU data, and onboard autonomy software. The buggy replaces manual steering with a motorized drive by wire system and uses a Jetson Orin Nano running ROS2 to process sensor data, estimate position, detect terrain features, and generate navigation commands.

Problem & solution

• Built an electric off road vehicle platform powered by a 48V 150Ah battery and high speed motor system
• Replaced traditional steering input with a motorized drive by wire control system for autonomous steering commands
• Integrated Pandar40P LiDAR, GPS, IMU, and onboard compute into a ROS2 based autonomy stack
• Used SLAM and sensor fusion to estimate vehicle position across rough outdoor terrain where GPS alone can be unreliable
• Designed control logic for speed, steering, obstacle awareness, route following, and safe manual override behavior
• Processed LiDAR point cloud data to support terrain mapping, obstacle detection, and local navigation decisions
• Structured the system around separate ROS2 nodes for sensor input, localization, planning, control, telemetry, and safety monitoring

What I learned

• How autonomous vehicles combine hardware, embedded systems, robotics software, and real time control
• How to use LiDAR, GPS, and IMU data together instead of relying on one sensor source
• How ROS2 organizes robotics systems into nodes, topics, messages, transforms, and launch files
• How drive by wire systems turn software commands into physical steering and motion control
• How to think about safety, fallback behavior, manual override, and testing when software controls a real vehicle
• How to debug robotics problems across sensors, wiring, Linux services, ROS2 logs, calibration files, and physical hardware