-
This repository provides C++ plugin files, and XML model and world files to control the operation of an autonomous vehicle in a simulated environment.
-
This repository is submitted for partial fulfillment of the Carleton College Computer Science department capstone seminar.
-
This project is an extension of a previous Capstone Seminar. The original project can be found at https://github.com/kileymaki/Self-Driving-Comps
Kirsten Baker, Tom Choi, Will Richards, Ruyi Shen, and Ben Withbroe
- Accurately detect and follow straight and curved lanes
- Detect and avoid reasonably sized obstacles and straight, and slightly curved lanes
- Get back to original lane after performing an avoidance maneuver.
- Transition between any of these operations on a closed course indefinitely.
High Level Controller
-Controls the high level decision making of the vehicle. Decides what ‘high level’ operation to perform ( avoid an obstacle, follow the road, return to lane etc.)
Low Level Controller
-controls the mechanical operation of the vehicle. Once the HLC ‘decides where to go’, the LLC ensures that the car will go there.
Lidar
-processes lidar sensor information from the cars lidar sensor to detect obstacles
Camera
-processes camera sensor information to detect road lanes, and generate waypoints along the road to follow.
GPS
-gets cars position in the world
Gazebo
- Physics simulator supporting our model, world, and plugin files -http://gazebosim.org
FADBAD++
- Gazebo dependancy for automatic differentiation - http://www.imm.dtu.dk/~kajm/FADBAD/
BOOST
- Gazebo dependancy
OpenCV
- Camera Vision mathematics library used in image processing - http://opencv.org
(1) brew install gazebo6 (We do not guarantee support for any versions of gazebo later than gazebo 6.0)
-this will also install dependant fadbad++ and boost library packages
(2) brew install opencv
- Install Dependancy
(3) git clone https://github.com/shenruyi1994/SelfDrivingCarTeamZero.git
-Clone our repo
(4) cd SelfDrivingCarTeamZero.git
-cd into cloned repo
(5) mkdir build
-Create build directory to hold our makefile
(6) cd build && cmake .. && cd ..
-Use CMake to generate makefile
(7) ./sdcMake
-Run script to build the project
(8) gazebo [worldfile.world] [-u]
- Pass gazebo the name of your worldfile as first paramater
- Use -u parameter to start the simulation in a paused state
Gazebo can be inconsistent in how it loads and renders worlds, so if something doesn't look right try running the .world file again.
If some or all of the models still aren't showing up correctly, copy the directories within the models directory in the repo to the models directory in your .gazebo directory. This is usually located in your home directory.
Starting the simulation in a paused state often helps with consistency as well.