8 ms·

Slightly off topic, but still some questions that have been bothering me for a while: LiDAR for autonomous vehicles is cool and all, but how will this perform

by zevv 8y ago

Slightly off topic, but still some questions that have been bothering me for a while: LiDAR for autonomous vehicles is cool and all, but how will this perform if there is not the occasional car making test runs through town, but when all vehicles on a busy road or intersection are equipped with a one or more units? I understand that both radar and ultrasonic suffer from interference, so will this be different? Are there plans for doing any kind of (cooperative?) frequency or time division to avoid interference between cars? How DOS-able would LiDAR be? Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer?

donkeyd 8y ago
> Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? Hang it under a drone and bring down air traffic at the same time!
baybal2 8y ago
> How DOS-able would LiDAR be? Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? Even that will not be necessary, a splat of dirt on the sensor is enough. I don't understand their obsession with LADARs. MM wave radars were successfully handling the same task in the industry for a few decades while being much cheaper and reliable. NIH too strong. I feel calling that here is 100% appropriate and objective.
- TomMarius 8y ago
  Is there anything I could buy and try that uses the technology you suggest?
  - baybal2 8y ago
    Quite a few Volvo's and some S Class Benz cars with distronic plus cruise control. Apparently also some Nissan cars, but only as an emergency braking sensor.
    - TomMarius 8y ago
      Yeah I have used the Mercedes system. Does it have the same obstacle detection capabilities as LIDAR?
      - baybal2 8y ago
        The brains processing the output certainly don't do that (emergency break logic is much simpler than automatic obstacle avoidance,) but nobody prevents making them do so by throwing more processing power on that.
    - late2part 8y ago
      I was surprised that the Toyota Camry I rented from Avis had it too.
- Symmetry 8y ago
  Radar is very useful but doesn't replace a lidar. Radars give very accurate readings of an objects distance and relative velocity but very poor angular resolution. For that reason they aren't much use for avoiding stationary objects since those are hard to distinguish from the ground, signposts, etc. At least until they get so close that the radar field of view lets you rule out those things. They are really great at figuring out whether cars are braking, though, since cars are moving with respect to the ground you can just filter the ground out and with radar you can sense the change in speed directly rather than having to figure out the speed by comparing distance over time. EDIT: Actually, I think automotive radars don't give any directional information at all since they don't use big dishes. I'm used to working with aerospace radars but even with those I wouldn't rely on one for car navigation.
  - baybal2 8y ago
    > Radar is very useful but doesn't replace a lidar. It does, and quite well. Without any additional processing, you easily get few cm accuracy, and sub-wavelength resolution is possible if you do. That's more than enough. Latest automotive radars use electronic beam steering – not much different from what is used in latest missiles. I'd say, companies opting for ladars for driving assists don't have good engineering expertise.
    - Symmetry 8y ago
      The range resolution in cm is very good but the angular resolution in radians basically doesn't exist.
kken 8y ago
> but how will this perform if there is not the occasional car making test runs through town, but when all vehicles on a busy road or intersection are equipped This is called "jamming" and was a real problem for early radar solutions. Curiously you well rarely see anyone mention this in the LIDAR space. For many of startups adressing this issue seems to be an afterthough. One proper way to solve this is to use orthogonal codes in the scanning scheme. There are different ways to implement this. The complexity depends a lot on the used LIDAR architecture. It would not be surprising to see some of the startup being weeded out by that during real-world tests.
- Animats 8y ago
  I've argued that your outgoing beam timing should have a few microseconds of random jitter, so that no jammer can synchronize to it. Military radars do things like that.
Symmetry 8y ago
People don't worry about it because it just isn't a concern. The duty cycle of any particular lidar in any particular direction is very low and so for any two lidars they will have their beams pointed at the same place very infrequently. It happens but in order to, say, drive in the rain self driving cars need to be able to reject a certain amount of sensor noise. I've had a gagggle of robots using planar lidar, where the risk of interference is much higher, all work together in the same room together and not noticed any degradation in performance. The reasons lidars are so useful it that they're very directional in a way that radar isn't. That means that an effective lidar jammer can't just emit enough to interfere when the lidar is pointed at it, it needs to cause every other object present to also emit that strongly. This is possible but will cause nearby people to catch fire, at which point it probably doesn't make sense to call the object a "lidar jammer".
mrfusion 8y ago
Traffic is already very DOS-able from a box of nails.
gruez 8y ago
>How DOS-able would LiDAR be? Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? You can already "jam" human eyes pretty easily with an "eye jammer" aka. 5 mW green laser pointer.
dllu 8y ago
Yes, lidars have interference, but much less than radar and sonar. From my other comment [0]: > Compared to other sensors, lidars are not that prone to interference because: > * it only takes 1 microsecond to make a ranging measurement up to 150 m, so your detector is on for a short time > * lasers only illuminate a small spot, and the detector is also looking at a similarly small spot, so it is unlikely for two lidars to point in the same spot > Now, even if it does interfere, you may see a stream of random points pointed towards the interference source. This may happen if, say, you point a lidar directly at the sun, or if you have multiple lidars mounted on the same vehicle. Such random points are easily rejected as outliers and do not affect the vast majority of the scene. Most self driving cars (I hope) should have outlier rejection schemes that deal with outliers caused by this and other sources, such as snow, smoke, and so on. Now, as for bringing traffic to a halt by bringing your portable lidar jammer... https://www.xkcd.com/1958/ https://www.xkcd.com/1958/ [0] https://news.ycombinator.com/item?id=18444775 https://news.ycombinator.com/item?id=18444775

by zevv 8y ago

donkeyd 8y ago

> Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? Hang it under a drone and bring down air traffic at the same time!

baybal2 8y ago

> How DOS-able would LiDAR be? Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? Even that will not be necessary, a splat of dirt on the sensor is enough. I don't understand their obsession with LADARs. MM wave radars were successfully handling the same task in the industry for a few decades while being much cheaper and reliable. NIH too strong. I feel calling that here is 100% appropriate and objective.

TomMarius 8y ago
Is there anything I could buy and try that uses the technology you suggest?
- baybal2 8y ago
  Quite a few Volvo's and some S Class Benz cars with distronic plus cruise control. Apparently also some Nissan cars, but only as an emergency braking sensor.
  - TomMarius 8y ago
    Yeah I have used the Mercedes system. Does it have the same obstacle detection capabilities as LIDAR?
    - baybal2 8y ago
      The brains processing the output certainly don't do that (emergency break logic is much simpler than automatic obstacle avoidance,) but nobody prevents making them do so by throwing more processing power on that.
  - late2part 8y ago
    I was surprised that the Toyota Camry I rented from Avis had it too.
Symmetry 8y ago
Radar is very useful but doesn't replace a lidar. Radars give very accurate readings of an objects distance and relative velocity but very poor angular resolution. For that reason they aren't much use for avoiding stationary objects since those are hard to distinguish from the ground, signposts, etc. At least until they get so close that the radar field of view lets you rule out those things. They are really great at figuring out whether cars are braking, though, since cars are moving with respect to the ground you can just filter the ground out and with radar you can sense the change in speed directly rather than having to figure out the speed by comparing distance over time. EDIT: Actually, I think automotive radars don't give any directional information at all since they don't use big dishes. I'm used to working with aerospace radars but even with those I wouldn't rely on one for car navigation.
- baybal2 8y ago
  > Radar is very useful but doesn't replace a lidar. It does, and quite well. Without any additional processing, you easily get few cm accuracy, and sub-wavelength resolution is possible if you do. That's more than enough. Latest automotive radars use electronic beam steering – not much different from what is used in latest missiles. I'd say, companies opting for ladars for driving assists don't have good engineering expertise.
  - Symmetry 8y ago
    The range resolution in cm is very good but the angular resolution in radians basically doesn't exist.

kken 8y ago

> but how will this perform if there is not the occasional car making test runs through town, but when all vehicles on a busy road or intersection are equipped This is called "jamming" and was a real problem for early radar solutions. Curiously you well rarely see anyone mention this in the LIDAR space. For many of startups adressing this issue seems to be an afterthough. One proper way to solve this is to use orthogonal codes in the scanning scheme. There are different ways to implement this. The complexity depends a lot on the used LIDAR architecture. It would not be surprising to see some of the startup being weeded out by that during real-world tests.

Animats 8y ago
I've argued that your outgoing beam timing should have a few microseconds of random jitter, so that no jammer can synchronize to it. Military radars do things like that.

Symmetry 8y ago

People don't worry about it because it just isn't a concern. The duty cycle of any particular lidar in any particular direction is very low and so for any two lidars they will have their beams pointed at the same place very infrequently. It happens but in order to, say, drive in the rain self driving cars need to be able to reject a certain amount of sensor noise. I've had a gagggle of robots using planar lidar, where the risk of interference is much higher, all work together in the same room together and not noticed any degradation in performance. The reasons lidars are so useful it that they're very directional in a way that radar isn't. That means that an effective lidar jammer can't just emit enough to interfere when the lidar is pointed at it, it needs to cause every other object present to also emit that strongly. This is possible but will cause nearby people to catch fire, at which point it probably doesn't make sense to call the object a "lidar jammer".

mrfusion 8y ago

Traffic is already very DOS-able from a box of nails.

gruez 8y ago

>How DOS-able would LiDAR be? Would I be able to bring traffic downtown LA to a halt in 2040 by bringing my portable LiDAR jammer? You can already "jam" human eyes pretty easily with an "eye jammer" aka. 5 mW green laser pointer.

dllu 8y ago

Yes, lidars have interference, but much less than radar and sonar. From my other comment [0]: > Compared to other sensors, lidars are not that prone to interference because: > * it only takes 1 microsecond to make a ranging measurement up to 150 m, so your detector is on for a short time > * lasers only illuminate a small spot, and the detector is also looking at a similarly small spot, so it is unlikely for two lidars to point in the same spot > Now, even if it does interfere, you may see a stream of random points pointed towards the interference source. This may happen if, say, you point a lidar directly at the sun, or if you have multiple lidars mounted on the same vehicle. Such random points are easily rejected as outliers and do not affect the vast majority of the scene. Most self driving cars (I hope) should have outlier rejection schemes that deal with outliers caused by this and other sources, such as snow, smoke, and so on. Now, as for bringing traffic to a halt by bringing your portable lidar jammer... https://www.xkcd.com/1958/ https://www.xkcd.com/1958/ [0] https://news.ycombinator.com/item?id=18444775 https://news.ycombinator.com/item?id=18444775