Route-Time Conflict Maps for Cooperative End-to-End Autonomous Driving
Route-Time Conflict Maps for Cooperative End-to-End Autonomous Driving
This research presents When2Go, a cooperative end-to-end driving system designed to share intent rather than just perception data, enabling vehicles to resolve conflicts under SAE J3016 Class B or higher automation. By broadcasting LiDAR detections with forecast motion and intended trajectories, the system constructs route-time conflict maps where an order gate network decides precedence based on right-of-way rules. Trained by imitating a rule-based expert from V2Xverse with relabeling for unobservable stops, When2Go achieves state-of-the-art performance on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving scores on InterDrive (without and with traffic), 70.0 on the public V2Xverse protocol, and 88.2 on MDrive, with 95.0% of vehicles remaining collision-free during evaluation.
The paper introduces When2Go, a cooperative end-to-end driving system that shares intent to resolve conflicts under SAE J3016 Class B or higher automation, achieving state-of-the-art scores on three closed-loop CARLA benchmarks while explicitly noting that benchmark runs used a PID controller for steering and disabled forced pushing of long-waiting vehicles.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
driving score
97.8
View original evidence
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5driving score
90.7
View original evidence
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5driving score
90.1
View original evidence
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5driving score
78.9
View original evidence
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5driving score
70
View original evidence
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2611.13494v1 Announce Type: new Abstract: Existing cooperative end-to-end driving works at Class A (status-sharing) of the SAE J3216 classes of cooperative driving automation (CDA): vehicles exchange what they perceive, not what they intend, so each plans alone when to go, which limits cooperation and leads to deadlock when both vehicles yield or to collisions when both go. We present When2Go, a cooperative end
Open source S2
s not learn to wait without an observable cause. For our runs only, the benchmarks' PID controller steers by When2Go's waypoints instead of the route, and its forced push of long-waiting vehicles is off. With one model, When2Go sets the state of the art on three closed-loop CARLA benchmarks: 97.8 and 90.7 driving score on InterDrive without and with traffic (best published 90.1 and 78.9), 70.0 on the public V2Xverse
Open source S5
protocol (best of sixteen rerun methods 59.1) and 88.2 on MDrive (best rerun 85.1), with 95.0% of its vehicles collision-free (87.7%).
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org