1. BaSaMa & HiWi
    2.  

Risk-Aware Motion Prediction in Autonomous Racing Using Particle Sampling

Institut
Lehrstuhl für Fahrzeugtechnik
Typ
Semesterarbeit / Masterarbeit /
Inhalt
experimentell / theoretisch /  
Beschreibung

Autonomous racing contributes to the challenge of making autonomous vehicles safer and more reliable in high-speed environments. One key factor for avoiding collisions but enforce overtakes is the motion prediction of racing opponents. To choose safer trajectories, when necessary, prediction algorithms could determine the probability of each motion prediction and evaluate the risk of interactions between the ego vehicle and opponents in autonomous racing.

The aim of this thesis is to develop and implement a risk-aware motion prediction system for autonomous driving using particle sampling. This method involves generating multiple predictions by varying parameters or employing different prediction approaches around the object of interest. The distribution of these predicted trajectories allows us to assess the likelihood of various outcomes and evaluate the associated risks in vehicle interactions. This approach enhances the ability of autonomous racing systems to make informed decisions under uncertainty, ultimately improving safety and performance on the track.

Work packages:

  • Literature review on the state of the art and methodologies related to motion prediction and risk assessment in autonomous driving
  • Design and Implementation of Particle Sampling Method and integration into current software stack
  • Analyze the distribution of the generated trajectories to determine the probability of each potential motion and develop algorithms to evaluate the risk
  • Evaluate the system's performance in predicting and mitigating risks in vehicle interactions
  • Discuss the findings from the evaluation phase, highlighting the strengths and limitations of the risk-aware motion prediction system

 

Voraussetzungen
  • independently familiarize yourself with the topic
  • creativity
  • a structured way of working
  • knowledge of ROS2
  • programming experience in C++
Sprache
deutsch / englisch
Tags
FTM Studienarbeit, FTM AV, FTM Esser, FTM Informatik
Möglicher Beginn
sofort
Kontakt
Daniel Esser, M.Sc.
Raum: MW 3505
Tel.: +49 89 289 15342
daniel.essertum.de
Ausschreibung