Skip to content

Add WarpDriver operational model#1545

Draft
chraibi wants to merge 7 commits intomasterfrom
001-warp-driver-model
Draft

Add WarpDriver operational model#1545
chraibi wants to merge 7 commits intomasterfrom
001-warp-driver-model

Conversation

@chraibi
Copy link
Contributor

@chraibi chraibi commented Mar 14, 2026

Summary

  • Add WarpDriver operational model (Wolinski, Lin, Pettré 2016) across all three layers: C++ core, pybind11 bindings, Python wrapper
  • Probabilistic collision field model: agents project trajectories, evaluate collision probability via warped intrinsic fields, and adjust velocity through gradient descent
  • Random lateral perturbation on trajectory samples to break symmetry in head-on encounters (speed model, no repulsion forces)
  • Boundary steering to keep agents away from walls
  • Jam detection with chill mode to prevent oscillation in dense crowds

Preview: https://PedestrianDynamics.github.io/jupedsim/pull-requests/1545/

chraibi added 2 commits March 14, 2026 09:37
Implement probabilistic collision avoidance based on Wolinski, Lin,
and Pettré (2016). The model uses warped intrinsic fields to predict
collisions and gradient descent to compute avoidance velocities.

Full three-layer integration: C++ core, pybind11 bindings, Python wrapper.
Includes system tests, bidirectional benchmark, and documentation.
- Remove agent-agent repulsion forces from WarpDriverModel (speed model,
  not force model)
- Add small random lateral perturbation on trajectory samples to break
  symmetry in head-on encounters
- Rename boundary repulsion to boundary steering for consistency
- Move WarpDriver docs into pedestrian_models/, add short description
  in index.rst with link to full model page
- Document agent steering as inactive option for future use
github-actions bot added a commit that referenced this pull request Mar 14, 2026
@chraibi chraibi marked this pull request as draft March 14, 2026 09:38
github-actions bot added a commit that referenced this pull request Mar 14, 2026
@chraibi
Copy link
Contributor Author

chraibi commented Mar 14, 2026

benchmark_models

@chraibi
Copy link
Contributor Author

chraibi commented Mar 14, 2026

Although the mode calculates fewer steps, it's overall a bit slower most other models.
Depending on the scenario, this might or might not hurt.

In this PR the model will be introduced. Optimised, if needed, in another step.

image

github-actions bot added a commit that referenced this pull request Mar 14, 2026
github-actions bot added a commit that referenced this pull request Mar 14, 2026
github-actions bot added a commit that referenced this pull request Mar 14, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant