LiDAR-Inertial Navigation Companion Integration
Budget: ₹750 – ₹1,250 INR
I need to fuse my LiDAR and IMU data on an off-board companion computer so the vehicle can build an accurate map of industrial boilers and keep its pose estimate rock-solid while navigating inside them. The sole focus is high-accuracy positioning; real-time processing speed and terrain adaptation are secondary nice-to-haves, but the map must not drift more than ±5 cm during a full inspection run.
Current stack
• 360° mechanical LiDAR and MEMS IMU already mounted
• Jetson Xavier NX running Ubuntu 22.04 + ROS 2 Humble
• Flight controller communicating through MAVLink for off-board control
What I need from you
• ROS 2 nodes (C++ or Python) or a complete package set that performs LiDAR-Inertial odometry/SLAM, tuned for the confined, reflective interior of boilers
• Launch, config, and calibration files specific to this sensor pair
• Clean MAVROS topic bridge so the flight controller receives continuous, filtered pose updates
• Step-by-step documentation plus a repeatable test procedure; include a sample rosbag proving the accuracy target inside a mock boiler environment
Acceptance criteria
• ≤ 5 cm cumulative drift over 10 minutes of continuous movement inside a boiler
• CPU load below 70 % on the Xavier NX during operation
• Code builds with colcon, uses only standard ROS 2 dependencies, and runs headless
You are free to base the solution on FAST-LIO2, LOAM, RTAB-Map, or a custom EKF/UKF—just explain your choice and show how it meets the accuracy requirement.
Share any sensor specs or clarifications you need, and we can dive right in.
Current stack
• 360° mechanical LiDAR and MEMS IMU already mounted
• Jetson Xavier NX running Ubuntu 22.04 + ROS 2 Humble
• Flight controller communicating through MAVLink for off-board control
What I need from you
• ROS 2 nodes (C++ or Python) or a complete package set that performs LiDAR-Inertial odometry/SLAM, tuned for the confined, reflective interior of boilers
• Launch, config, and calibration files specific to this sensor pair
• Clean MAVROS topic bridge so the flight controller receives continuous, filtered pose updates
• Step-by-step documentation plus a repeatable test procedure; include a sample rosbag proving the accuracy target inside a mock boiler environment
Acceptance criteria
• ≤ 5 cm cumulative drift over 10 minutes of continuous movement inside a boiler
• CPU load below 70 % on the Xavier NX during operation
• Code builds with colcon, uses only standard ROS 2 dependencies, and runs headless
You are free to base the solution on FAST-LIO2, LOAM, RTAB-Map, or a custom EKF/UKF—just explain your choice and show how it meets the accuracy requirement.
Share any sensor specs or clarifications you need, and we can dive right in.