search

autonomousvision / transfuser

[PAMI'23] TransFuser: Imitation with Transformer-Based Sensor Fusion for Autonomous Driving; [CVPR'21] Multi-Modal Fusion Transformer for End-to-End Autonomous Driving
MIT License
1.1k stars 184 forks source link

How to measure the target_vehicle_distance from label_raw.json #234

Closed SRajasekar333 closed 1 month ago

SRajasekar333 commented 1 month ago

Hi,

I am trying to extract the target_vehicle_distanceform thelabel_raw.json, but I am facing some issues to exactly find the target_vehicle_hazard from the car_id list, so please guide me on how to define the exact target vehicle hazard.

From label_raw.json, we could find that the "Car_id" which has the "distance": -1 to be as the current ego_vehicle. and from measurements.json we could find the vehicle_hazard: [True, False] .

Now how to find the "target_Car_id" from the label list which will be the potential target_vehicle_hazard (for e.g.. vehicle which lies in the same lane of the ego_vehicle and which lies infront of the ego_vehicle to be called as a potential hazard) and then to extract the corresponding "distance": of the car_id

Thanks!

Kait0 commented 1 month ago

We don't store the information about which vehicle caused the hazard flag from the expert. You could collect a new dataset and store that information. The autopilot does not really have a notion of target_vehicle.