Hi everyone,
I’d like to share our recent progress on the ROBOSN BO1 Quester, an industrial inspection drone platform we’ve been developing based on the PX4 ecosystem.
Our main goal was to solve the stability issues in complex environments where GPS is unreliable or unavailable. We’ve done extensive field testing involving Lidar-Visual fusion, precision landing, and night operations.
Here are some of our latest flight test results. We’d love to hear your feedback on the flight performance!
1. The Toughest Challenge: GPS-Denied Forest Navigation
This is our most recent test in a dense forest. As many of you know, flying autonomously here is tricky due to light changes and unstructured obstacles. We utilized Lidar-Visual fusion to achieve stable localization.
Tech Note: Notice the stability during the transition into the tree canopy. The EKF fusion logic was tuned to prioritize local position estimates here.
2. Indoor-Outdoor Transition & Autonomous Charging
We also tested the seamless transition from indoor to outdoor environments and the integration with an automatic charging stock.
3. Low-Light Performance: Night Inspection
Visual positioning usually fails at night. We tested our sensor fusion performance in a low-light environment (7:18 PM) to ensure 24/7 operation capability.
4. Putting it all together: The Full Inspection Workflow
Finally, here is a demo of a complete automated workflow, simulating a real-world industrial inspection mission from takeoff to data capture.
Summary: Building this on top of PX4 has been an exciting journey. We are still optimizing the obstacle avoidance algorithms and the path planning efficiency.
If you have any questions about our setup or the tuning process, feel free to ask!