Skip to content

mbalabanski/learned-dynamics

Repository files navigation

Trajectory Tracing

Learned Dynamics

Learned vehicle dynamics project for AE 4968 with DCSL.

Try in colab

Usage

This project uses UV build system. Install uv with pip install uv

Clone repository and download dependencies using uv sync.
Note that PyTorch requires a specific CUDA version -- this project uses PyTorch with CUDA 13.0. Edit the PyProject.toml, UV sources section to fetch a different CUDA version.

To run the streamlit app, run:

uv run streamlit run quick_plot_app.py

This should launch a web app where you will select a model and verify its rollouts.

You should see a figure resembling the vehicle dynamics stochastic process, with a side-by-side of the Pacejka dynamics model integrated with Euler's method (green) and the spatiotemporal forecasting of the model (blue).

Sampling Stochastic Process

To train a new model, use the script train.py

About

AE 4698 Vehicle Dynamics Project

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages