Simulation Subsystem

Overview

The Simulation Subsystem takes care of simulating the realistic environment given model from the Modelling subsystem. Its major function are running a specific scenario and supporting multiple agents simultaneously. This system also deals with integration with Steering wheel systems like Logitech G920.  We will primarily be using CARLA software as the simulation platform.

System Implementation

System diagram for running a scenario in simulation:

We will be utilizing and extending CARLA‘s s repository called Scenario Runner for running a specific scenario. This will generate a scenario based on a configuration file. The custom agent developed by us from the modelling system will be then started and assigned a goal position to achieve. Everything together will serve as a realistic environment.

As the baseline model, we are currently using the “Learning by Cheating” model.

 

Targeted Requirements

  • M.P. 6 Have a simulation speed of minimum 10 FPS supporting at least 3 simultaneous agents.

 

Current Status

  • Baseline behavioral model working with 3 agents within the range of 9-11 server FPS.
  • Support to run specific scenarios using baseline behavioral model 

 

Future work

  • To optimize the performance of simulation with S.T.A.R.S. behavior model, 
    • Client-side development by introducing a simulator bridge
    • Server-side development by adding code for running the model in the simulator itself. 

Strong/Weak Points

  • The current system performs as per the requirements. However, it is at the border of the fps values required. This provides almost no room to accommodate any new development to improve the model.  
  • On the bright side, the analysis made on the codebase has provided some insights regarding how this can further be improved. As we have already developed a framework that is working, we can confidently spend time improving performance (fps).  
  • To support different CARLA maps retraining is required.

Results

We can see one such scenario playing out here with our baseline model.