Skip to content
This repository has been archived by the owner on Nov 16, 2023. It is now read-only.

feat(#157): Started lidar mapping #160

Draft
wants to merge 10 commits into
base: main
Choose a base branch
from

Conversation

timdreier
Copy link
Contributor

Description

Map the lidar points to the front RGB image, to get the depth (distance) of segments later.

Fixes #157

Time invested

Tim Dreier: 8h (so far)

Type of change

Please delete options that are not relevant.

  • New feature (non-breaking change which adds functionality)

Does this PR introduce a breaking change?

No.

Most important changes

That it's possible to get the depth of any segment in a image.

Checklist:

  • My code follows the style guidelines of this project
  • I have performed a self-review of my own code
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works (might be obsolete with CI later on)
  • New and existing unit tests pass locally with my changes (might be obsolete with CI later on)

@timdreier timdreier linked an issue Jan 24, 2023 that may be closed by this pull request
@timdreier timdreier marked this pull request as draft January 24, 2023 20:50
@github-actions
Copy link

Simulation results

Metric Value
Avg. driving score 0.069615
Avg. route completion 0.255
Avg. infraction penalty 0.273
Collisions with pedestrians 0.0
Collisions with vehicles 54.151
Collisions with layout 54.151
Red lights infractions 0.0
Stop sign infractions 0.0
Off-road infractions 0
Route deviations 0.0
Route timeouts 54.151
Agent blocked 0.0
Yield emergency vehicles infractions 0.0
Scenario timeouts 54.151
Min speed infractions 0.0

@timdreier
Copy link
Contributor Author

Additional time for progress: ~12h

@nylser nylser self-assigned this Mar 1, 2023
@timdreier
Copy link
Contributor Author

timdreier commented Mar 21, 2023

Additional effort to implement Buffer node and debugging: 4 hours

@timdreier
Copy link
Contributor Author

The most important thing in this PR is the fork of lidar-camera (https://github.com/timdreier/lidar-camera). Since we hadn't enough time we couldn't fix all errors regarding this node.

What works: We get a PointCloud filtered by the viewport of the according camera:

Bildschirm­foto 2023-03-25 um 14 02 39

The points should have the color they have in the camera image, what obviously doesn't work.

Furthermore, the node creates a depth map. Sadly, the depth map is missing some points. This could be improved by optimizing the LIDAR-Parameters to send more LIDAR points in the Viewport of the Camera. The current output is like this:

Bildschirm­foto 2023-03-25 um 14 07 13

Furthermore, this PR contains a buffer node, which basically just overrides Depth-Values in the Depthmap if they are greater than a threshold (5000 in this example). The hoped to get a more consistent image by doing so. However, this could cause some problems because the depthmap may contain old values.

would be interesting to test the state things in future work!

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[Feature]: map lidar points to rgb image
2 participants