FSSIM

FSSIM is a vehicle simulator dedicated for Formula Student Driverless Competition. It was developed for autonomous software testing purposes and not for gaming. A version of this simulator was used to predict lap time of gotthard at FSG 2018 trackdrive with 1% accuracy.

This simulator is developed and tested on Ubuntu 16.04 and ROS Kinetic and both are assumed to be already installed.

The more extensive tutorial can be found under Wiki

FSSIM is developed by Juraj Kabzan as part of our work at AMZ-Driverless.

How to Run It in your Workspace

Install sudo apt install ros-kinetic-desktop-full and sudo apt install python-catkin-tools
Clone this repository to an existing ROS Workspace initialized with catkin init
Run cd src/fssim from the workspace.
Run ./update_dependencies.sh, you will need to approve multiple packages to be installed
Run catkin build
Source the workspace
After successful building, run the simulator with roslaunch fssim auto_fssim.launch. RVIZ window will start. NOTE: You might need to untick and tick FSSIM Track and RobotModel in RVIZ in order to load the STL files. NOTE: This [Wrn] [ModelDatabase.cc:339] Getting models from[http://gazebosim.org/models/]. This may take a few seconds. might take up to a minute when starting for the first time.
The terminal will inform you what is happening. The loading time takes around 20 seconds. When Sending RES GO will show up in the terminal, you can start controlling the vehicle with /fssim/cmd topic.

Combine it with simple FSD skeleton Framework and drive a lap

Install sudo apt install ros-kinetic-desktop-full and sudo apt install python-catkin-tools
Clone the AMZ skeleton workspace.
Run ./update_dependencies.sh -f from fsd_skeleton, you will need to approve multiple packages to be installed
Compile with catkin build
Source the workspace source fsd_environment.sh
Run roslaunch fssim_interface fssim.launch
Run roslaunch control_meta trackdrive.launch

Run automated test. Execute FSD_ATS (this command is loaded when sourcing fsd_environment.sh). If you will want to see the visualization, run RViZ: roslaunch fssim_interface rviz.launch. NOTE: The car will keep driving until the simulation will time-out since no lap counter is implemented.

Features

This simulator is targeted for FSD competition, thus it contains some of the real-car safety features
- RES (Remote Emergency Stop): The vehicle will not be able to be controlled if a /fssim/res_state/push_button = true is not sent. On the other side, if /fssim/res_state/emergency = true is send, the vehicle will stop immediately. If you start FSSIM with roslaunch fssim auto_fssim.launch or through fssim_interface this is done automatically.
- Leaving Track: If the simulation is started with auto_fssim.launch, an automated RES person is launched. This means, if the vehicle exists the track with all four wheels, RES-emergency will be send and the simulation will exit itself
FSSIM does not simulate the RAW sensors! It uses a cone-sensor-model instead. This means a cone observations around the vehicle are simulated with numerous noise-models. The configuration file for this sensors can be found in fssim/fssim_description/cars/gotthard/config/sensors.yaml. Thanks to this simplification it is real-time capable
FSSIM does not use GAZEBO Physics Engine to simulate the vehicle. Instead, it uses a basic vehicle model which is discretized with Euler Forward discretization and overwrites the model pose. This feature allows the simulated model to match closely the real world car.
Automated Run of multiple scenarios with different configurations and save logs as well as report.
Lap-time counter

Known problems

The update rate of TF between wheels and main chassis is too low or too rough. This causes jumps of wheels at higher speeds in RViZ. This influences only the visualization and not the functionality.

Cite

If using this project for scientific publications, please cite this paper:

Juraj Kabzan, Miguel de la Iglesia Valls, Victor Reijgwart, Hubertus Franciscus Cornelis Hendrikx, Claas Ehmke, Manish Prajapat, Andreas Bühler, Nikhil Gosala, Mehak Gupta, Ramya Sivanesan, Ankit Dhall, Eugenio Chisari, Napat Karnchanachari, Sonja Brits, Manuel Dangel, Inkyu Sa, Renaud Dubé, Abel Gawel, Mark Pfeiffer, Alexander Liniger, John Lygeros, Roland Siegwart, "AMZ Driverless: The Full Autonomous Racing System", arXiv preprint arXiv:1905.05150

@misc{1905.05150,
Author = {Juraj Kabzan and Miguel de la Iglesia Valls and Victor Reijgwart and Hubertus Franciscus Cornelis Hendrikx and Claas Ehmke and Manish Prajapat and Andreas Bühler and Nikhil Gosala and Mehak Gupta and Ramya Sivanesan and Ankit Dhall and Eugenio Chisari and Napat Karnchanachari and Sonja Brits and Manuel Dangel and Inkyu Sa and Renaud Dubé and Abel Gawel and Mark Pfeiffer and Alexander Liniger and John Lygeros and Roland Siegwart},
Title = {AMZ Driverless: The Full Autonomous Racing System},
Year = {2019},
Eprint = {arXiv:1905.05150},
}

Example

A note: this is a public copy of a private version. The public version might have some internal functionality removed. FSSIM was developed by

AMZ-Racing / fssim

readme