Master’s Thesis - RSU-based Cooperative Perception using BEVFusion and ETSI ITS Communication

22.05.2026, Diplomarbeiten, Bachelor- und Masterarbeiten

The Chair of Robotics, Artificial Intelligence, and Real-Time Systems offers a Master’s thesis focusing on extending an existing CARLA-based cooperative perception framework with Roadside Units (RSUs). The thesis investigates how infrastructure-based sensing can improve uncertainty-aware cooperative perception using BEVFusion and ETSI ITS communication standards.

Motivation & Relevance

Urban intersections remain one of the most challenging environments for autonomous vehicles due to occlusions, limited visibility, and dynamic traffic participants. Cooperative perception systems can improve scene understanding by sharing information between vehicles and roadside infrastructure.

An existing research pipeline already supports CARLA-based multi-agent simulation, BEVFusion-based perception, and uncertainty-aware object detection. However, the current system primarily focuses on vehicle-centric perception. Extending the framework with RSU-based sensing enables the investigation of infrastructure-assisted cooperative perception and visibility enhancement in occluded urban scenarios.

This thesis shall build upon the following repositories:

• ETSI ITS ROS2 Messages: https://github.com/ika-rwth-aachen/etsi_its_messages
• Shared Local Dynamic Map (S-LDM): https://github.com/DriveX-devs/S-LDM/tree/main

Project Description

In this thesis, you will extend the existing cooperative perception framework with roadside infrastructure sensors and ETSI ITS communication support.

The system currently supports:

• CARLA-based cooperative perception
• BEVFusion-based sensor fusion
• Uncertainty-aware object detection
• Multi-agent simulation workflows

Your task is to integrate RSU-based perception into the pipeline and enable the exchange of cooperative perception information using CPM and CAM messages.

Your system will:

• Integrate one or multiple RSU camera systems into CARLA
• Implement perception modules for RSU-based object detection
• Generate CPM messages from RSU perception outputs
• Fuse RSU and vehicle-based perception information
• Evaluate awareness improvements in occluded intersection scenarios

The result will be an infrastructure-assisted cooperative perception framework for urban autonomous driving.

Your tasks

• Extend the CARLA cooperative perception pipeline with RSUs
• Develop RSU-based perception modules
• Implement ROS2-based CPM generation
• Fuse RSU and vehicle perception outputs
• Evaluate uncertainty-aware cooperative perception performance

Your Profile

• Master’s student in Computer Science, Robotics, Electrical Engineering or related field
• Proficiency in Python or C++
• Interest in computer vision and autonomous driving
• Experience with ROS2, CARLA, or perception systems is beneficial

What you will gain

• Expertise in cooperative perception systems
• Hands-on experience with BEVFusion and uncertainty-aware perception
• Experience with ROS2 and ETSI ITS communication
• Insight into intelligent transportation systems

How to apply

Please send your CV and a transcript of your grades with your application.

Kontakt: erik-leo.hass@tum.de