NGSIM
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Trajectory Prediction | NGSIM | ADE (Avg)0.77 | 26 | |
| Vehicle Trajectory Prediction | NGSIM (test) | RMSE0.2733 | 25 | |
| Traffic State Estimation | NGSIM Experiment Set 2 | MAE1.041 | 24 | |
| Trajectory Prediction | NGSIM | RMSE (1s)0.3 | 18 | |
| Trajectory Prediction | NGSIM | Precision99.89 | 18 | |
| Car-following acceleration prediction | NGSIM 80/20 vehicle split Pipeline S (test) | R2 Score97.2 | 11 | |
| Trajectory Prediction | NGSIM | RMSE (1s)0.23 | 11 | |
| Trajectory Prediction | NGSIM | ADE2.31 | 11 | |
| Car-following acceleration prediction | NGSIM Pipeline R rolling-mean filtered, 0.8 s shift (test) | Rolling R237.54 | 10 | |
| Lane Keeping and Lane Change Classification | NGSIM I-80 | Accuracy96.98 | 8 | |
| Highway on-ramp merging | NGSIM Hard Traffic Scenario | Success Rate99.8 | 7 | |
| Highway on-ramp merging | NGSIM Medium Traffic Scenario | Success Rate99.87 | 7 | |
| Highway on-ramp merging | NGSIM Easy Traffic Scenario | Success Rate99.6 | 7 | |
| Vehicle Path Prediction | NGSIM | RMSE (1s)0.15 | 7 | |
| Lane-change prediction | NGSIM I-80 strict vehicle-ID holdout (502 unseen drivers) (test) | Balanced Accuracy67.4 | 6 | |
| OOD Detection | NGSIM Highway | AUROC93 | 6 | |
| Trajectory Prediction | NGSIM 5 sec. | ADE0.4 | 6 | |
| Open-loop simulation | NGSIM (100 held-out vehicles) | Crash Rate0.01 | 5 | |
| Trajectory Prediction | NGSIM | E2E Latency (Jetson Nano 10W)27.87 | 5 | |
| Motion Planning | NGSIM Highway Cut-in Scenario | Pc (%)0 | 5 | |
| Density Reconstruction | NGSIM Four-Loop Detector | MAE (veh/m)5.01 | 4 | |
| Speed Reconstruction | NGSIM Four-Loop Detector | MAE (m/s)4.57 | 4 | |
| Density Reconstruction | NGSIM 40% Penetration Rate (test) | MAE3.2 | 4 | |
| Speed Reconstruction | NGSIM 40% Penetration Rate (test) | MAE2.53 | 4 | |
| Density Reconstruction | NGSIM 30% Penetration Rate (test) | MAE3.37 | 4 |
Showing 25 of 38 rows