Re: [sumo-user] Reinforcement Learning in Matlab

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

- if a vehicle controlled by moveToXY reaches the end of it's edge, it exits the simulation unless  either of the following is done

  - the moveToXY command is sent in every simulation step

  - the vehicle was provided with a route that continues after the edge

- once the vehicle has left the simulation, you can no longer control it

- the error given by sumo corresponds to the TraCIException that you catch in you control script. It communicates to you that you TraCI command could not be executed as intended

Am Fr., 11. Feb. 2022 um 17:46 Uhr schrieb lllcccmmm--- via sumo-user <sumo-user@xxxxxxxxxxx>:

Hello, I'm coding an RL agent in Matlab to decide the best actions at each junction for an autonomous vehicle moving into a SUMO road map. The first training simulation goes well. After the first simulation the vehicle goes outside the map and the instruction traci.vehicle.getRoadID(veh_ID) gives my an error that interrupt the entire project. To overcome this issue I introduce a Matlab try/catch. In this way, Matlab produce the final output. My problem is that SUMO says "Error: Answered with error to command 0xa4: Vehicle 'vehicle_0' is not known." and the RL training breaks.  Nevertheless Matlab doesn't present any error.

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user