GEM Workspace and Starter Code
This starter code implements a baseline for lane detection and lane following control.
Do note:
For information on e.g., what rostopics to record, software versions, hardware specs, etc. PLEASE, please look at the vehicle documentation, linked on the course project page and directly in the CfA resources here.
When recording rosbags, you can change the buffer size if you get warnings about packets dropping; read the cli.
The majority of the code that you will be working with in this starting code can be found at the path e#/src/vehicle_drivers/gem_gnss_control/scripts. The README within that specific directory is tailored for explaining the usage of the gem_gnss_tracker_pid.py+ lane_detection.py system.
If you want to use the baseline lane following, you will need to download the corresponding model weights as a .pt file. You can read the README.md at the directory mentioned in the line above for instruction on this.
The ROS environment for e2 vs e4 is very similar. There are just a few small differences, e.g., 4 arena (corner) cameras on e4 versus a front-left front-right only corner cameras on e2. The actual python code (not various CMake configs), is nearly identical, aside from potential some parameters about vehicle length/width, etc.
You are not constrained to using a PID controller (as is done in gem_gnss_tracker_pid.py for lane following). We have also implemented Pure Pursuit and Stanley controllers in gem_gnss_tracker_pp.py and gem_gnss_tracker_stanley_rtk.py respectively. However, note that the Pure Pursuit and Stanley controllers do not currently support integration with the lane detection module / lane following. You would have to do some integration work (though, that should not be too hard).
You can choose to only use pieces of this starter code (or, you can choose to use none of it). Purely up to you what you want to do