README.md

Overall System Architecture

This is meant to be run in combination with the ForzaETH race_stack

Base System: Either the Simulator or the physical system. They are both made to work interchangeably with each other i.e. offer the same outputs/inputs to the outside, as well as the same functionality. The base system launches many different nodes that originate inside the utils repo as well.

• State Estimation
• Localisation
• Mapping
• Sensor drivers
• Odometry
• Motor Control via VESC
• Car State

Utils: Glues Simulator, physical system, racestack and the pit together. If there is cross repo functionality required, it belongs into utils s.t. interchangeability between the base systems is guaranteed.

• Frenet Conversion Service
• Frenet State Republisher
• Overtaking Sectors
• Speed Sectors
• Tuner Packages
• Custom Messages
• etc

Race Stack: Here is the high level autonomy stack i.e. Perception, Planning and Control for simplicity we use a state machine as well.

• Perception
• Planning
• Control
• State Machine

For quick usage refer to the passages below:

Tele-Operation

To simply run the car in tele-op mode, i.e. remote controlled driving, you can use the following command, where NAME_OF_RACECAR is the name of the racecar you are using in the stack_master/config (e.g. JET1):

roslaunch racecar pbl_teleop.launch racecar_version:=NAME_OF_RACECAR

Hint: Make sure your controller is turned on! This is only meant to be used on the physical system

Mapping

MAP_NAME is the name you want to give to the map that you are now mapping. It will save the map in the stack_master/maps folder here. The default map that exists for the simulator is called f. You can use maps which were mapped in reality seamlessly in the simulator as well, but not vice-versa, as the physical system needs the .pbstream for the SLAM algorithm to localise.

Physical System

roslaunch stack_master mapping.launch racecar_version:=NAME_OF_CAR map_name:=MAP_NAME

Simulator

roslaunch stack_master mapping.launch map_name:=MAP_NAME racecar_version:=SIM sim:=True

Note: You NEED to map first also in simulation.

Base System

In the base system you can choose which localisation algorithm you want to use, either slam Cartographer or synpf SynPF.

Physical System

roslaunch stack_master base_system.launch map_name:=MAP_NAME racecar_version:=NAME_OF_CAR algo:=<slam or synpf>

Simulator

roslaunch stack_master base_system.launch map_name:=MAP_NAME sim:=True racecar_version:=<SIM or NAME_OF_CAR> tire_model:=<linear/pacejka>

The simulator will use the parameters identified with the system identification procedures. You can find the source parameters here

Time Trials

You can choose the control algorithm to be used, either MAP or PP (Pure Pursuit). If you use PP, you do not need to specify the LUT.

Physical System

roslaunch stack_master time_trials.launch LU_table:=NAME_OF_CAR_hangar_pacejka ctrl_algo:=<MAP or PP>

Simulator

roslaunch stack_master time_trials.launch LU_table:=SIM_linear ctrl_algo:=<MAP or PP>

Head to Head

You can choose the control algorithm to be used, either MAP or PP (Pure Pursuit). If you use PP, you do not need to specify the LUT. The overtaking mode can be set to spliner as of know and future overtaking algorithms will be added.

Physical System

roslaunch stack_master headtohead.launch LU_table:=NAME_OF_CAR_hangar_pacejka ctrl_algo:=<MAP or PP> overtake_mode:=spliner

Simulator

roslaunch stack_master headtohead.launch LU_table:=SIM_linear ctrl_algo:=<MAP or PP> overtake_mode:=spliner

To test head-to-head in sim, run the following to publish a dummy opponent:

roslaunch obstacle_publisher obstacle_publisher.launch speed:=1

This builds on top of the original F1TENTH System