Competitions and Challenges:

• https://carlachallenge.org/challenge/
• NeurIPS challenges

What is Motion Planning problem?

Motion planning

Given the observations and the predictions of the dynamic actors around the self-driving vehicle, as well as the static constructs provided by HD maps, the problem of motion planning is to efficiently solve for the trajectory the self-driving vehicle will follow.

Which simulator and framework to use for training procedures in learning-based approaches?

• SMARTS: https://github.com/huawei-noah/SMARTS (Scalable Multi-Agent RL Training School for Autonomous Driving)
• SUMO-RL: https://github.com/LucasAlegre/sumo-rl https://github.com/eclipse/sumo
• CARLA: Example of how to work with RL and CARLA: https://www.youtube.com/watch?v=MynOPIJi4VY (Gym with CARLA: https://github.com/PacktPublishing/Hands-On-Intelligent-Agents-with-OpenAI-Gym/tree/master/ch8/environment) https://github.com/Sentdex/Carla-RL https://www.youtube.com/watch?v=MynOPIJi4VY
• L5Kit + DriverGym: https://woven-planet.github.io/l5kit/ https://github.com/woven-planet/l5kit (Python library with functionality for the development and training of learned prediction, planning and simulation models for autonomous driving applications.) (https://woven-planet.github.io/l5kit/reinforcement.html)
• https://github.com/eleurent/highway-env: A minimalist environment for decision-making in autonomous driving
• https://github.com/Stanford-ILIAD/CARLO
• https://github.com/learn-to-race/l2r RL + AD for racing https://learn-to-race.org/ based on Arrival simulator with gym interface (High fidelity)
• https://github.com/vista-simulator/vista RL +AD
• https://github.com/praveen-palanisamy/macad-gym MARL built based on CARLA
• SUMMIT
• https://deepdrive.io/index.html Deepdrive Unreal engine
• https://github.com/flow-project/flow Traffic Mircosimulation with DRL