ParkSim

Vehicle simualtion and behavior prediction in parking lots. This is a monorepo with different projects mixed-in.

Authors: Xu Shen (xu_shen@berkeley.edu), Alex Wong, Neelay Velingker, Matthew Lacayo, Nidhir Guggilla

ParkPredict+

ParkPredict+: Multimodal Intent and Motion Prediction for Vehicles in Parking Lots with CNN and Transformer

Authors: Xu Shen, Matthew Lacayo, Nidhir Guggilla, Francesco Borrelli

Install

1. Clone this repo
2. In the setup.py script, enable the following packages
   • tensorboard
   • pytorch_lightning
   • einops
3. In the /python folder of this repo, do pip install -e . (A virtualenv is recommended)
4. If components of DLP dataset is needed, install the DLP package and request data according to the instructions there.
5. Install pip install the correct pytorch version into this virtualenv based on your OS and hardware.

Usage

1. A pre-trained intent prediction model can be downloaded here.
2. A pre-trained trajectory prediction model can be downloaded here.

Use this notebook to evaluate the pre-trained models.

See this page for information about training.

Fleet Parking

Parking of Connected Automated Vehicles: Vehicle Control, Parking Assignment, and Multi-agent Simulation

Authors: Xu Shen, Yongkeun Choi, Alex Wong, Francesco Borrelli, Scott Moura, Soomin Woo

Configuration

File Preparation

1. In your BASE folder, clone this repository
2. Go back to the BASE level, clone the DLP dataset repository
3. Now your folder structure should look be:

BASE
    – ParkSim
    – dlp_dataset

4. Extract the attachment zip from this link
5. Create a folder dlp-dataset/data, move the files DJI_0015_*.json into it, i.e.

BASE
    – ParkSim
    – dlp_dataset
        — data
            – DJI_0015_*.json

6. Create a folder ParkSim/data, move the rest of the files into it, i.e.

BASE
    – ParkSim
        – data
            – other_files_in_the_zip
    – dlp_dataset
        – data
            – DJI_0015_*.json

Install

1. create a virtual environment and activate it
2. Enter the ParkSim/python folder
3. Run pip install -e . to install parksim package
4. With the current virtualenv activated, go into the dlp-dataset folder and run pip install -e . to install the dlp package into the same env
5. Install pip install the correct pytorch version into this virtualenv based on your OS and hardware.

Test

1. With the virtualenv activated, run python python/parksim/simulator/rule_based_simulator.py to run the simulation
2. You can change different simulation settings with this yaml file

Note: If you are testing within a WSL2 system with Windows built-in WSLg, and cannot see any GUI after simulation starts, try setting export LIBGL_ALWAYS_SOFTWARE=1 before running it.

Note: The ROS-related code is out-of-sync with the project development.