DriveVT

This project demonstrates a real-time, on-road, lattice-based autonomous vehicle motion planner in the browser. Many autonomous vehicle motion planners are implemented close to the metal in C or C++, or they utilize computing platforms like CUDA or OpenCL to generate plans in a highly parallel fashion on the GPU. Using WebGL, we can implement similar parallel planning algorithms right in the browser that can run on a variety of moderately-powerful consumer graphics cards. With Three.js, the motion planner can be executed in real-time 3D simulated scenarios.

This simulator is still a work-in-progress. You may see bugs, WebGL crashes, strange vehicle behavior, and any number of other failures. Currently, it is fully functional only in Google Chrome with hardware acceleration enabled and experimental canvas features enabled. The target frame rate is 60fps, but it may run slower depending on how long planning takes on your hardware.

Next steps: try out several example scenarios or create your own.

The source code and more information can be found in the README.

© 2018 Matt Bradley • MIT License • GitHub