CommonRoad is an open-source software focuses on motion planning problem, the traffic scenarios in CommonRoad can convert from the real-traffic datasets (such as NGSIM, highD, inD…). Many researchers developed platoon toolbox based on open-source software with the development of CACC technology, the problem of multi-vehicle interaction still has strong research value, CommonRoad-MARL provides the engine dynamics model and platooning, by adding the engine dynamics model, one can get more realistic dynamic behavior and this also can be used for heterogenous platooning research.
In real-traffic and platooning, vehicle(agent) needs to interact with other participants or other vehicles in platoon. For this interactive environment, multi-agent reinforcement learning (MARL) is strongly correlated. With MARL, all vehicles in platooning have more intelligent and safe potential, which also consistent with the trend of future development，so we developed CommonRoad-MARL to support Multi-agent reinforcement learning research for platooning.
