ROSA - The Roundabout Optimized Speed Advisory

This repository contains the official implementation for the paper: “ROSA: Roundabout Optimized Speed Advisory with Multi-Agent Trajectory Prediction in Multimodal Traffic”, to be presented at IEEE ITSC 2025.

ROSA combines multi-agent trajectory prediction with coordinated speed guidance for vehicles approaching and entering roundabouts in multimodal and mixed traffic environments. It supports both automated and human-driven vehicles and provides proactive, real-time speed advisories based on predicted conflicts with other vehicles and Vulnerable Road Users (VRUs).

Installation

Prerequisites

• Anaconda or Miniconda

Cloning the repository

git clone git@github.com:urbanAIthi/ROSA.git
cd ROSA

Setting up the Conda environment

Use the following command to create a Conda environment based on the environment.yaml file:

conda env create -f environment.yaml --name rosa
conda activate rosa

Installing SUMO

This repository requires that you already have installed the SUMO traffic simulator. For more information on how to install SUMO, please refer [to https://sumo.dlr.de/docs/Installing/index.html]. This project uses SUMO version 1.22.0. Please be aware that simulation results may vary when using different SUMO versions due to changes or updates in the simulator.

Downloading dataset

The openDD trajectory dataset (rdb1) is used for this work: [https://l3pilot.eu/data/opendd.html]. To use the model with the rdb1 scenario, download the dataset from this link and ensure that trajectories_rdb1_v3.sqlite is located in the ROSA_pred/rdb1 directory.

Project Structure

This repository is organized into two main components, reflecting the architecture of the ROSA system as described in the paper.

1. ROSA_pred/ – Transformer-Based Multi-Agent Trajectory Prediction

This module implements a Transformer-based model that jointly predicts the trajectories of all agents (vehicles and VRUs) at roundabouts.

• Model architecture: Transformer Encoder with attention masking to capture agent dynamics and inter-agent dependencies.
• Input features: Position, velocity, acceleration (lateral/tangential), heading angle (sin/cos), agent type, and optional exit intention.
• Training: Supervised single-step prediction based on 3-second history.
• Inference: Autoregressive prediction over a 5-second horizon.
• Output: Predicted trajectories including kinematic state (position, speed, acceleration, orientation).
• Evaluation: ADE/FDE and binary classification for occupancy of crosswalks and entries.

How to run

cd ROSA_pred
python main.py

The prediction settings (model parameters, training options, etc.) can be modified in train_configs.yaml.

Structure

ROSA_pred/  
  ├── configs/ # dir containing config files  
  ├── rdb1/ # dir containing roundabout specific files like dataset and shapefiles from openDD  
  ├── trained_models/ # dir containing models trained for the ROSA paper
  ├── utils/ # dir containing different utility functions  
  ├── generate_dataset.py # file for dataset preprocessing  
  ├── models.py # file for model handling  
  ├── main.py # main file for starting the preprocessing-training-testing pipeline
  ├── trainer.py # file for training the prediction model
  ├── tester.py # file for testing the prediction model

2. ROSA_sim/ – Speed Advisory and SUMO Simulation

This module implements the ROSA speed advisory algorithm and evaluates it using realistic roundabout scenarios modeled in SUMO (Simulation of Urban MObility).

• Goal: Improve safety and efficiency by proactively adjusting vehicle speed based on predicted crosswalk and entry occupancy (derived from the ROSA trajectory prediction module).
• Functionality: Model-based calculation of optimal speed.
• Evaluation: 6,600 scenarios from the openDD dataset modeled in SUMO, considering the following metrics: fuel/energy consumption, emissions, travel time, waiting time, and stops.

How to run

cd ROSA_sim
python run.py

The simulation settings (scenarios, vehicle type, etc.) can be modified in config.ini.

Structure

ROSA_sim/  
  ├── env/ # dir containing simulation environment files (evaluation, speed advisory)
  ├── sumo/ # dir containing sumo configuration files
  ├── utils/ # dir containing different utility functions  
  ├── config.ini # file for simulation settings
  ├── run.py # main file for starting the simulation

Citation

If you use this code in your research, please cite our paper:

@INPROCEEDINGS{ROSA_Schlamp,
  author={Schlamp, Anna-Lena and Gerner, Jeremias and Bogenberger, Klaus and Huber, Werner and Schmidtner, Stefanie},
  booktitle={2025 IEEE International Conference on Intelligent Transportation Systems (ITSC)}, 
  title={ROSA: Roundabout Optimized Speed Advisory with Multi-Agent Trajectory Prediction in Multimodal Traffic}, 
  year={2025},
  volume={},
  number={},
  pages={},
  doi={}
}

The Transformer implementation is based on the following repository [https://github.com/urbanAIthi/FloatingCarObserver] and the associated papers:

@INPROCEEDINGS{10422398,
   author={Gerner, Jeremias and Rößle, Dominik and Cremers, Daniel and Bogenberger, Klaus and Schön, Torsten and Schmidtner, Stefanie},
   booktitle={2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)}, 
   title={Enhancing Realistic Floating Car Observers in Microscopic Traffic Simulation}, 
   year={2023},
   pages={2396-2403},
   doi={10.1109/ITSC57777.2023.10422398}
}

@article{gerner2025FCO_TFCO,
  author       = {Jeremias Gerner and Klaus Bogenberger and Stefanie Schmidtner},
  title        = {Floating Car Observers in Intelligent Transportation Systems: Detection Modeling and Temporal Insights},
  journal      = {arXiv},
  year         = {2025},
}

License

This project is licensed under the MIT License. See the LICENSE file for details.