Task Force - Real-World Autonomous Driving: Object Tracking and Motion Prediction in Urban and Racing Application

Institut

Lehrstuhl für Fahrzeugtechnik

Typ

Semesterarbeit /

Inhalt

experimentell / theoretisch /  

Beschreibung

To advance the field of autonomous driving (AD), we leverage our research platforms - EDGAR (Figure 1, left) and TUM Autonomous Motorsport, TAM (Figure 1, right) - to develop state-of-the-art AD software. While urban and racing AD application domains differ significantly, the symbiotic relationship between these platforms allows us to transfer independent learnings, enhancing our overall AD capabilities.

This IDP project, composed of 4 individual topics, aims to establish a robust evaluation benchmark for AD perception software across both urban and racing domains, and to implement state-of-the-art (sota) algorithms on EDGAR and TAM. Our primary focus is on perception, the critical discipline of understanding a vehicle's environment in the present and future. Perception encompasses three key tasks: detection, tracking, and prediction.

Recent research underscores the importance of improved tracking and prediction, revealing their potential to deliver substantial performance gains. Tracking enables the association of objects over time, providing a foundation for accurate prediction, which involves forecasting the motion of agents into the future.

Through this project, we aim to:

1. Develop a modular pipeline to evaluate and benchmark tracking and prediction algorithms for use on EDGAR and TAM.
2. Implement sota tracking and prediction algorithms on EDGAR for real-world applications.

These efforts will help identify potential improvements in tracking and prediction, pushing the limits of autonomous driving performance in both urban and racing scenarios.

 

Keywords: Deep Learning, Multi-object Tracking, Motion Prediction, Autonomous Driving, Autonomous Racing, EDGAR, TAM

If you are interested in any of these topics, please apply for a specific one or inform us of your preferences.

Voraussetzungen

Requirements:
- Good programming skills in Python or C++
- Ideally experience with ROS2
- High personal motivation and independent working style
- Very good language proficiency in German, English, or French

Software

Python, C++, ROS2

Tags

FTM AV, FTM AV Perception, FTM Stratil, FTM Informatik

Möglicher Beginn

sofort

Kontakt

Loïc Stratil, M.Sc.
Raum: MW 3508
Tel.: +49.89.289.15898
loic.stratiltum.de

Ausschreibung