At Terabee we did some intensive experiments with TeraRanger Tower and Pixhawk running ArduCopter. The purpose of the experiment was to explore the obstacle avoidance indoors using rangefinders only.
In our setup we used TeraRanger Tower with 8 sensors (in a separate test we used 4 sensors in a + configuration. It worked flawlessly) and a single TeraRanger Evo looking down as an altimeter (without GPS the obstacle avoidance works in altitude hold mode).
One of the things we learned is that AVOID_DIST_MAX parameter has the minimum value of 3.0 meters for a reason. When we forced the value to 1.5 meters we crashed into a wall, twice. Our best guess is that the aircraft didn’t have enough time and control authority to move away from the wall (we allowed the obstacle avoidance to request up to 5 degrees lean angle).
We are really happy with the results of our experiments and we are really excited to see ArduPilot growing further. We would like to thank the ArduPilot dev team for their fabulous work!
If you would like to know more about our testing of indoor avoidance, here is a much longer writeup containing all the technical details I could think of.
If you have any questions about our setup then I’m always happy to help!
Comments
@Chris Anderson - I don't think it compares. In the object avoidance mode only lean angles are commanded to move away from obstacles, we don't have any information about the position. With Pozyx or Marvelmind we should be able to get quite precise position information and use some of the modes that operate on groundspeed / groundacceleration.
If you meant adding Pozyx or Marvelmind to this setup then we would be curious how it behaves in loiter mode. Maybe we will try this as a future experiment.
Very impressive! How do you think this compares to Pozyx and Marvelmind in the same application?