SMARTS (Scalable Multi-Agent RL Training School)
2022 NeurIPS Driving SMARTS Competition
End Notice
The 2022 NeurIPS Driving SMARTS Competition has now ended!
Many thanks to all the organizers and participants who contributed to the competition.
The competition page can be seen at 2022 Neurips Driving SMARTS Competition
Results
The competition objective is to train a single policy capable of controlling single-agent or multi-agent to complete different driving tasks in various scenarios.
Track 1
Task: Any method may be used to develop the policy.
Rank | Prize | Team | Solution | Presentation |
---|---|---|---|---|
1st | 6000 USD | tjudrllab-fanta | fanta-code | tjudrllab_fanta_driving_smarts |
2nd | 4000 USD | VCR | SMARTS_VCR | vcr_driving_smarts |
3rd | 2000 USD | AID | Predictive-Decision | aid_driving_smarts |
Congratulations to the Track 1 winners!
Track 2
Task: An offline learning method using offline datasets must be used to develop the policy.
Rank | Prize | Team | Solution | Presentation |
---|---|---|---|---|
1st | 6000 USD | tjudrllab-fanta | fanta-code | tjudrllab_fanta_driving_smarts |
2nd | 4000 USD | VCR | SMARTS_VCR | vcr_driving_smarts |
3rd | 2000 USD | AID | Predictive-Decision | aid_driving_smarts |
Congratulations to the Track 2 winners!
Bonus Awards
Category | Prize | Team |
---|---|---|
Most Innovative | 1000 USD | AID |
Participation (Random selection) | 1000 USD | tjudrllab-mb |
Congratulations to AID for being unanimously voted by the organizing team as having the most innovative solution!