LiDAR implementation

[LucaTrs97]

Hi, I was interested in simulating a 360° LiDAR sensor with stonefish but I noticed that it doesn't exist yet.

For the meantime, I'm simulating the LiDAR output using 4 depth cameras with 90 deg FoV. Is there any plan to implement it? Or do you think there is a better / smarter idea to simulate the LiDAR with the already existing tools? The depth camera approach kinda works but there is some distortion in the point cloud generate by the depth cameras.

Thanks in advance
Luca

[patrykcieslak]

Hello!
I was not implementing a Lidar as I was focusing on the underwater sensors but there is a Multibeam2 sensor that I think was basically a Lidar.

There is no other efficient way to simulate it than using depth cameras and to correct the distortion you need to remember that the angle is not changing linearly with the pixel position along the frame so some remapping has to be done. I do it for the sonars. Using 90 deg fov cameras is too much - you get a significant difference in spatial resolution across the frame.
I will look into the Multibeam2 and probably remake it to be a Lidar because it makes not much sense as an acoustic sensor.

[Mariolopez31]

Hello!
I’ve been working on this exact issue and I wanted to share that I've developed a fork of Stonefish and stonefish_ros2 with two LiDAR implementations:
- A standard 360° LiDAR: For general use.
- Livox Mid-360: Specifically mimicking its scanning pattern.

Following Patryk's advice, I used the Multibeam2 sensor as a base. I’ve already tested it with algorithms like FAST-LIO and the results look solid!
It’s still a "v1", so there's definitely room for improvement (especially regarding performance), but it's a functional starting point.