LiDAR simulation - contraflow traffic detection

[mc_axe]

Hi there, long time

I hope everyone is doing fine,

We are working on a DYI device, that aims to detect contraflow traffic, by taking distance measurements and throwing alerts, on a real street (We is a club of tech enthousiasts),
Assume we have 1 or more LiDAR "eyes", placed on a street, constantly taking measurements with 500Hz, pointed like a concentrated beam (with really small divergance 12mRad) towars traffic (divergance is optional)

How can i get the distances of moving subjects (meshes in our case), on a table? For small Animations tops 10 or 20sec,
I was thinking that will be easier to animate various scenarios in a program like SI,
(From one moving vehicle -> ideal scenario, to multiple that also change lanes or accel/decel Etc),

Rather than writing a top 2D view simulation from scratch that will be a simplified version (boxes) at best,
3D measurements might help us deside the nr of eyes their positioning (orientation, placement) Etc, and save us from un-ending discussions, and manually drawing triangles and circles like Archimedes on the ground (while remembering very little trigonometry)

Any suggestions? What comes first in your mind ICE or a script?
Hint: Havent touched softimage for i while,

The goal is to minimize real world testing, and simmulate measurements to develop the program that will take the dessisions,
We havent got permission yet but we are preety enthousiastic about it, we plan a fast track r&d because the problem is very real where i live,

Also wanted to test if there is any pulse in here
Thanks in advance <3

[FXDude]

Hi!

You could investigate ice raycasting, there are a number of videos from pooby,
but the first one should cover the basics ...



That video was to represent scalar values on the raycasting grid based on distance,

but you could cast rays from a vehicle, and with some "conditional logic"
if greater than, or lesser than a certain distance from other vehicles
can trigger different "events" like "slowing down" or "accellerating" or "changing lanes".

So let us know if you want to try something, or if running into "roadblocks" ( pun intended )

[mc_axe]

Thanks FXDudue this will help alot!




The goal is to simmulate "recorded" data, no need to apply conditional logic in here,

I also logged the values in the script editor,
I might set the scene to 500fps, since the lidars have that frequency,

Awesome stuff,
We need to start from ideal values also there is no noise involved in here,

But i might add some later, to make it more challenging like the real world measurements, or even try to implement divergance :o

But y thats a nice start, thanks again

[FXDude]

Huh! cool ! : )

and Wow --> 500fps lol

Otherwise, with all the "self driving" talk, I hope "autonomous vehicles" would all be "independent" ...
or would not require a net connection that would 3d scan everywhere we drive-by, to be sent to car manufacturers,
and/or otherwise tracking every single car ...

.. and in that regard, what you're doing is a step ( or a drive ) in the right direction !

( also what do you mean by "divergeance" ? )

[mc_axe]

Oh sorry, i spelled that wrong, my bad,

Its Divergence,

For a circular beam, divergence is defined as the angular measure of how the beam diameter increases with the distance from the laser aperture

I was thinking that maybe i can raycast through a sligtly curved grid (or maybe through a thin slice of a sphere), to create a cone of rays
And then avg, the sum of measurements, comming from that beam, ,

That and some noise and i think we are ready to go,
-

I cant stress how much i agree, in regards to self driving vehicles, its a huge discussion, freedom vs control,
Maybe when AI models are trained to perfection, cars coudl think on their own, in a localized env, without sending all data back for more and more "training",
We might get in a point where that is a part of a premioum package for VIPs (that want to avoid tracking), and then masses might demand it as well, who knows?

Anyways practically, i believe cars (selfdriving or not) could use feedback from other cars, and other sensors,
Our sensors will NOT report all traffic (they could) but they will alert when contraflow traffic is detected.