When Does Physics Beat Data? Data-Efficient Traffic State Estimation with Amortized PINNs
Organizations: HSE University, Moscow, Russia · Moscow Institute of Physics and Technology (MIPT), Russia · AI VK, Moscow, Russia
Abstract
Traffic state estimation (TSE) reconstructs traffic density at unmonitored locations from a sparse set of sensors. Data-driven models work well when sensors are dense but degrade sharply as coverage thins. We show that imposing the Lighthill-Whitham-Richards conservation law as a physics-informed prior makes TSE more data-efficient. On four real corridors (NGSIM, I-80, PeMS04, PeMS08), physics-informed estimators generalise to held-out sensors where data-driven models, including classical and deep kriging, overfit the observed ones. On the two shock-rich corridors the effect is a clear crossover: as more sensors are held out, data-driven models collapse while the physics-informed estimator degrades gracefully, leading the no-physics network by about 26-38% at 85-90% held out and the strongest classical kriging by about 10-21% at 90%. A matched smoothness prior does not reproduce the gain, and an extended Kalman filter on the same LWR model is outperformed on the shock corridors, so the advantage comes from how the conservation law is imposed. Finally, one loss-conditional PINN removes the per-corridor loss-weight search, matching a 20-point search at about 5x lower cost.
Figures & tables
| Method | NGSIM | I-80 | PeMS04 | PeMS08 |
|---|---|---|---|---|
| LC-PINN (no weight search) | 0.262 | 0.248 | 0.444 | 0.406 |
| Tuned LWR-PINN | 0.229 | 0.226 | 0.487 | 0.450 |
| No-physics (data-only) | 0.203 | 0.198 | 0.537 | 0.544 |
| GP-kriging | 0.190 | 0.187 | 0.548 | 0.507 |
| NN-kriging | 0.192 | 0.182 | 0.582 | 0.486 |
| IGNNK (deep kriging) | 0.219 | 0.261 | 0.588 | 0.543 |
| Method | .30 | .50 | .70 | .85 | .90 | |
|---|---|---|---|---|---|---|
| NGSIM | LC-PINN | 0.262 | 0.271 | 0.278 | 0.302 | 0.351 |
| No-physics | 0.203 | 0.220 | 0.303 | 0.462 | 0.569 | |
| GP-kriging | 0.190 | 0.170 | 0.230 | 0.382 | 0.471 | |
| NN-krig | 0.192 | 0.197 | 0.251 | 0.354 | 0.445 | |
| IGNNK | 0.219 | 0.221 | 0.599 | 0.926 | 0.999 | |
| EKF (phys) | 0.488 | 0.496 | 0.538 | 0.598 | 0.661 |
Traffic Flow Estimation
Traffic flow estimation infers and predicts vehicle volumes, speeds, densities, and origin–destination demand from incomplete transportation data.
See all 56 papers