0
$\begingroup$

I will soon be working on an autonomous drone project and we're planning to use LIDAR systems for localization & mapping.

However, at the same time, we are looking into how to make drones safer when used in a crowded (human) environment. This also means taking vandalism into account.

  • Can Lidar sensors be manipulated/affected by people on the ground? If yes, how? (perhaps a laser-pointer?)
  • Since Lidars are reflecting light on surfaces, are there surfaces it cant detect, such as glass?
  • Would a Solid State Lidar be more reliable (prone to vibrations) for this kind of application?
$\endgroup$
0
$\begingroup$

Lidar systems are all going to be some wavelength of infrared, which means there isn't going to be a visible light component on the ground and, conversely, no amount of visible light emitted from the ground is really going to have any impact on your drone.

Specifically to your question, a visible laser being shot from the ground should have virtually no effect on the lidar unit. It might be possible to get an IR laser, but if that's the case then you're already dealing with someone more advanced than your run-of-the-mill vandal.

Once someone gets an IR laser, they'll then also need to be capable of spotting it in order to be able to target the drone. I've got first-hand experience trying to spot the lidar output from about 20 meters and I can tell you it's very difficult even with night-vision goggles, a dim viewing area, and knowing where the spot should be.

This means also that whoever is targeting your drone is going to need some very sophisticated IR/night vision equipment to spot the infrared laser from whatever distance you're going to be flying. Again, even at 20 meters, which isn't really that high, the laser spot is going to be very difficult to see, and again also they'll only really be able to interfere with your system at night.

So, from that end, I wouldn't really worry about someone being able interfere with drone.

Regarding your other points, yeah, the lidar needs the emitted laser light to be reflected back at the unit, so you're going to have difficulty measuring highly reflective surfaces and highly absorbent surfaces.

Finally, I would not be as concerned with the platform vibration affecting the scanner as the other way around - the lidar units have a spinning head inside that turns typically at 10-25 Hz or so, which means basically you're going to wind up rigidly coupling a gyroscope to your airframe. You may find this degrades your ability to control the quadcopter. You could of course get around this by using a solid-state lidar, but I've worked with quite a few lidar companies (SICK, Velodyne, Quanergy, Ouster) and nobody has even a demo unit for us to evaluate.

$\endgroup$
6
  • $\begingroup$ Many thanks! These are great insights and very helpful for our process. Can you recommend us a lightweight, middle-class LIDAR? We are planning to use a DJI Air 2S which can carry 450 - 750 grams $\endgroup$
    – Azrion
    Jul 1 at 9:33
  • $\begingroup$ I also read more about the Time of Flight (ToF) since most lidars work like this. And the biggest issue I see with it is, getting wrong information because the IR light is reflected multiple times, for example, by corners. Do you have a keyword for me that I can search for to minimize such errors? It is difficult to find information at all about LIDARS.. $\endgroup$
    – Azrion
    Jul 1 at 9:36
  • $\begingroup$ @Azrion - regarding search terms, I don't have one for you, but there's not going to be a lot you can do about this with just the lidar data alone, because the lidar believes the measurement to be true. I know SICK has multi-echo technology, so they can report the first 3 or 5 or whatever returns. Some scanners will only report the first return, which glitches on dust or rain/fog, some scanners only report the most intense return, which glitches on reflections, so it's really important to understand just what you're getting from the scanner. $\endgroup$
    – Chuck
    Jul 1 at 13:19
  • $\begingroup$ Lidar units are only going to give you a direction and a distance, and it's up to you to do some signal processing on that data. You don't know how many bounces the beam took, so it would be difficult or impossible to tell the difference between hitting a puddle in the street, bouncing, and reflecting off the side of a house and shooting into a storm drain and measuring into the sewer system. Both shots are going to look like a hole in the ground because the reflecting case is going to report the total distance (puddle + bounce) at the angle it left the scanner, ie to the street. $\endgroup$
    – Chuck
    Jul 1 at 13:22
  • $\begingroup$ Finally, regarding particular models, I would say you need to establish your needs first and then use that to filter models. Do you need one plane or 32? Is 8 okay? SICK's most famous model, I think, the LMS511, is relatively huge and overweight, but does great in severe locations. They have the TIM or TiM or whatever, but what do you need for range? Scan rate? Angular resolution? These are all questions you need to resolve before getting to the point of picking a particular model. Once you've got a few contenders, ask your sales rep for a demo/evaluation unit! $\endgroup$
    – Chuck
    Jul 1 at 13:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.