README.md

Motion Prediction

This guide explains how to set up and run motion prediction using the TAROT framework and the UniTraj repository.

Step-by-Step Instructions

1. Navigate to the motion prediction example directory
   cd examples/motion_prediction

2. Clone the UniTraj repository
   Ensure you have SSH access to GitHub before running the following command:
   git clone [email protected]:vita-epfl/UniTraj.git

3. Install necessary dependencies for UniTraj
   Follow the instructions in the UniTraj README to set up dependencies.

4. Move TAROT-specific scripts into the UniTraj directory
   Move the tarot_train.py and tarot_config.yaml files into the cloned UniTraj repository:

   mv tarot_train.py UniTraj/unitraj

   mv tarot_config.yaml UniTraj/unitraj

   Download pretrained Autobot Ckpt into UniTraj/unitraj/unitraj_ckpt

5. Navigate to the UniTraj directory
   cd UniTraj

6. Run the TAROT training script
   Use the following command to start the training process:
   python tarot_train.py

Notes

• Make sure to configure tarot_config.yaml according to your requirements before running the training script.
• For issues with dependencies or environment setup, refer to the UniTraj documentation or TAROT-specific documentation.

Example Directory Structure

After completing the steps, your directory structure should look like this:

examples/motion_prediction/
├── UniTraj/unitraj
│ ├── tarot_train.py
│ ├── tarot_config.yaml
│ └── ... (other UniTraj files)