Corvus One - ROS-based L4 Autonomy for indoor drones

Hi friends!

We’ve been building for a few years Corvus One, which we believe is the world’s first Level 4 Autonomous MAV for large-scale indoor environments. By L4 Autonomy we mean the UAV has to be completely autonomous in its entire mission cycle, without any human intervention or use of specialized localization infrastructure. It’s built on ROS and PX4.

(The drone must be unmanned in scheduled takeoff, global and local path planning, finding and going to objects of interest, collision avoidance, landing and recharging, data sharing, and the ability to do all of that repeatedly! :slight_smile: )

Check out our video (shot a few months ago:)

Happy Tuesday!

Jackie Wu
CEO, Corvus Robotics

14 Likes

This is super exciting. I’ve been sitting on this video for awhile. This is really great work!

2 Likes

This is cool, thanks for sharing @robotjackie!

Could you share a bit more about the ROS - PX4 interaction? Are you using ROS or ROS 2? What are the limitations/hurdles you are observing for achieving (your definition of) L4 in drones? Are the hurdles in the ROS side or on the Autopilot one?

2 Likes

An amazing project!! Indoor navigation has always been a challenge! I’d love to know more about the specifics of the implementation.

1 Like

Thanks everyone!!! @vmayoral My cofounder Kabir is giving a talk on this exact topic in a few months at ROScon, if accepted. :slight_smile: Should be interesting to see then!

Looks awesome, can you give some insight on hardware components? Is this a complete customized build or did you start off with an existing platform? The landing / automated recharging feature looks really neat. Also, did you use px4 integrated components of VIO / Obstacle Avoidance, etc. or are these mostly individual ROS nodes?

1 Like