Skip to content

uncap-project/uncap-project.github.io

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

61 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

UNCAP: Uncertainty-Guided Neurosymbolic Planning Using Natural Language Communication for Cooperative Autonomous Vehicles

Neel P. Bhatt* · Po-han Li* · Kushagra Gupta* · Rohan Siva
Daniel Milan · Alexander T. Hogue · Sandeep P. Chinchali · David Fridovich-Keil
Zhangyang Wang · Ufuk Topcu
The University of Texas at Austin
*Equal contribution
AAMAS 2026 (Oral)


TL;DR

UNCAP is a neurosymbolic planning framework for cooperative autonomous vehicles that uses natural language communication + uncertainty reasoning to improve multi-agent coordination under partial observability.

It enables vehicles to:

  • Communicate what they know (and don’t know) in natural language
  • Perform uncertainty-aware planning
  • Achieve more robust and safer cooperation in complex driving scenarios

Framework Overview

Framework Overview

Autonomous vehicles operating in multi-agent environments face two key challenges:

  1. Partial observability (each agent has limited perception)
  2. Uncertainty in other agents’ intentions and environment state

Key Idea

UNCAP introduces a neurosymbolic pipeline that combines:

  • Perception (deep models)
  • Symbolic reasoning (planning under uncertainty)
  • Natural language communication (LLMs)

Core Components

  • Uncertainty Estimation
    Quantifies confidence in perception and predictions.

  • Language-Based Communication
    Agents exchange structured natural language messages describing:

    • Observations
    • Uncertainty
    • Intentions
  • Neurosymbolic Planner
    Integrates communicated information into a joint planning process.

  • Cooperative Decision-Making
    Produces safer and more efficient multi-agent trajectories.


Setup

1. Install Dependencies

Follow installation for:

2. Environment Setup

conda activate opencood

pip install ultralytics
pip install openai
pip install python-dotenv

Demo Notebook in: OPV2V/run.ipynb

Citation

If you find this work interesting and use it in your research, please consider citing our paper.

@inproceedings{bhatt2025uncap,
            title={{UNCAP}: Uncertainty-Guided Planning Using Natural Language Communication for Cooperative Autonomous Vehicles},
            author={Neel P. Bhatt and Po-han Li and Kushagra Gupta and Rohan Siva and Daniel Milan and Alexander Todd Hogue and Sandeep P. Chinchali and David Fridovich-Keil and Zhangyang Wang and ufuk topcu},
            booktitle={The 25th International Conference on Autonomous Agents and Multi-Agent Systems},
            year={2025},
            url={https://openreview.net/forum?id=aYlKa5ppLh}
}

About

[AAMAS 2026 (Oral)] UNCAP: Uncertainty-Guided Planning Using Natural Language Communication for Cooperative Autonomous Vehicles

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors