Hidden Biases of End-to-End Driving Models

Open paper
Benchmark Model Rank Results
autonomous-driving-on-carla-leaderboardTF++ WP#3Driving Score: 66.32Route Completion: 78.57Infraction penalty: 0.84
bench2drive-on-bench2driveTransFuser++#3Driving Score: 84.21
carla-leaderboard-2-0-on-carlaTF++ (Map Track)#3Driving Score: 5.56Route Completion: 11.82Infraction Score: 0.47
carla-leaderboard-2-0-on-carlaTF++#4Driving Score: 5.18Route Completion: 11.34Infraction Score: 0.48
carla-longest6-on-carlaTransFuser++ WP (TF++WP)#1Driving Score: 73Route Completion: 97Infraction Score: 0.56
carla-longest6-on-carlaTransFuser++ (TF++)#3Driving Score: 69Route Completion: 94Infraction Score: 0.72
carla-map-leaderboard-on-carlaMap TF++#1Driving score: 61.17Route completion: 81.81Infraction penalty: 0.70

Rank counts only results with a code link.