Submission Instruction

Participants need to inference on test dataset and submit their result file to our portal.
Test Dataset has been released in this link
So the Final Submission Deadline is 23:59 (GMT+8), June 6, 2025.
Submissions received by this time will be considered for final ranking and awards. The leaderboard will remain publicly accessible and actively maintained after the deadline.

Input and Output Requirements

Input Constraints

Front Camera Images from Vehicle

Camera Images from infrastructure

Commands (turn left, go straight, or turn right)

Ego Status at Current Frame

Intrinsics and Extrinsics of Sensor

Output Constraints

Temporal perception results

End-to-End planning results

Output Format Constraints

The perception and planning results should be encapsulated in a pkl-format document. We provide an inference pipeline with official baseline here
Example: We provide a sample inference file on the val dataset here, structured as follows:

if you want to submit results for Track 1 and Track 2:

{
    results: [
        {
            token: str
            # Tracking results
            boxes_3d: LiDARInstance3DBoxes
            scores_3d: Tensor, shape=(N,)
            labels_3d: Tensor, shape=(N,)
            track_scores: Tensor, shape=(N,)
            track_ids: Tensor, shape=(N,)
            # Detection results
            boxes_3d_det: LiDARInstance3DBoxes
            scores_3d_det: Tensor, shape=(M,)
            labels_3d_det: Tensor, shape=(M,)
            # Planning Results (future 5 seconds 2Hz traj points)
            planning_traj: Tensor, shape=(1, 10, 2)
        },
        {……},
        ……
    ]
}
                  

if you only want to submit results for Track 1:

{
    results: [
        {
            token: str
            # Tracking results
            boxes_3d: LiDARInstance3DBoxes
            scores_3d: Tensor, shape=(N,)
            labels_3d: Tensor, shape=(N,)
            track_scores: Tensor, shape=(N,)
            track_ids: Tensor, shape=(N,)
            # Detection results
            boxes_3d_det: LiDARInstance3DBoxes
            scores_3d_det: Tensor, shape=(M,)
            labels_3d_det: Tensor, shape=(M,)
        },
        {……},
        ……
    ]
}
                  

if you only want to submit results for Track 2:

{
    results: [
        {
            token: str
            # Planning Results (future 5 seconds 2Hz traj points)
            planning_traj: Tensor, shape=(1, 10, 2)
        },
        {……},
        ……
    ]
}
                  

Method Illustration Document

Participants are encouraged to submit an illustration file, such as a published paper, an ArXiv paper, a paper submitted to our workshop, or other technical reports, to explain their excellent methodology.
The file should provide sufficient details and analysis. A lack of explanation may undermine the credibility of the results. Organizers may reach out to teams with insufficient documentation for further clarification.

Submission Portal

Please submit your pkl-format result and method illustration to this form.
You may edit and resubmit this table as needed. The final version we receive will be considered the official submission.
Each team should designate only one member to submit the entry.
Prohibition: 1. Submitting results that are entirely based on existing others' open-sourced projects without any innovation is strictly prohibited. 2. The use of additional training data is not allowed. 3. Historical ego status is not allowed in our challenge.