We are excited to announce that two papers from the TUM Autonomous Vehicle Systems Lab will be presented at ECCV 2026.

The European Conference on Computer Vision (ECCV) is one of the world’s leading scientific conferences in computer vision and artificial intelligence. It brings together researchers from academia and industry to present advances in visual perception, machine learning, autonomous systems, and related fields.

Our accepted papers focus on understanding and planning from visual observations for autonomous systems:

• EgoDyn-Bench — Evaluating Ego-Motion Understanding in Vision-Centric Foundation Models for Autonomous Driving
  Finn Schäfer
  EgoDyn-Bench investigates how well vision-centric foundation models understand the motion of the ego vehicle from video, providing a benchmark for evaluating dynamic driving understanding.
   
• Target-Bench — Mapless Planning with Semantic Targets in Video World Models
  Dingrui Wang
  Target-Bench explores planning with semantic targets in video world models, enabling embodied systems to reason about and navigate toward meaningful targets without relying on predefined maps.