This project contains code for visualizing unsafe and reachable state sets
in a simple adapative cruise control scenario. The src folder contains the MATLAB code to generate the data, along with the viz folder which
can be served with any basic HTTP server.
This was originally made for a graduate course in "Safe Autonomy for Cyber-Physical Systems." The parameterized visualization interface is meant to be friendly and informative to those without much prior knowledge in the relevant areas (differential equations, reachability theory, etc.), and so it includes an expandable introduction and guide. The plan is to improve and expand the interface then host it as a free educational tool.
The visualization can be hosted with any HTTP server, for example by
running python -m http.server in the src/viz/ directory. Instructions
for using the interface are self-contained.
The data for the visualization is already included in this repo (about 300MB),
but it
can be regenerated with the MATLAB file at src/data_generation.m.
This script writes to CSV files in various folders which need to be
collated together by running python csv_collater.py
in the src/viz/ directory.
The system dynamics for the adaptive cruise control scenario are based on the benchmark in the paper "Networked Cooperative Platoon of Vehicles for Testing Methods and Verification Tools" (Makhlouf and Kowalewski 2015).
Reachability calculations are done using the CORA toolbox for MATLAB.
The code submitted by the CORA team for the 2020 ARCH-COMP (for which the paper above provides one of the benchmarks) was referenced while learning to use CORA with systems defined by differential equations.
Visualizations are implemented with D3.js.