What is possible contingency for lidar error during terrain following?

If in the event that lidar may experience faulty equipment or getting error. I wonder any parameters that I can set to prevent Lidar sending the drone to space when terrain following mission?

Anyone has any idea?

This depends on the error, sufficiently insane data would be filtered by the EKF.
I am not sure how much smartness is used, if you added two lidars in the same direction (nadir?)
They must not interfere with eachother, but I do not know if EKF would determine the one giving incorrect data.

I have a ground lidar. I just wonder what would happen in the following scenarios, what is ardupilot response.

  1. Lidar failed in mid flight during terrain following.
  2. Lidar giving inaccurate data for instance 30metres is given only 3 metres.

Would the drone drop out of the sky? Would ardupilot switch to using other altimeter?

  1. am I able to switch back to absolute altitude in mid air if I detect such error?

1 if data is lost, (as if lidar dies, no connection on a digital link) - as oppsite to a analog lidar - it would just lose that source, I believe EKF would just use it’s other estimation. - but that should be tested - you can do it by switching off power to a I2C/Serial LIDAR device.

for #2 , the first “jump” from 30 to 3 will be filtered out for a while, then if target is 30mAGL and reported is 3mAGL - it would climb …and continue to climb…

certainly not just “drop out of the sky”
#3 - yes, you can change EKF parameters midflight, and also switch mode to stop/pause mission while doing so.

the way a digital lidar works, makes it very unlikely that it will fail with a sane value (that requires a timely return signal) - it’s far more likely that a obscured/fogged/dirty sensor will report a preset “error value” which is higher that what you devine as maximum lidar range. If you define 120m , and Lidar reports any higher number, like the no-return indicating “130” - then AP will simply not use that data.(only log it)
I have seen many SF11C 's “get stuck” (report same value for several seconds - like 18.2m … this was fast ,high-vibration fixed-wing vehicles) Autopilot was excellent at not being fooled by that.

2 Likes