Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Overcoming Blind Spots: Occlusion Considerations for Improved Autonomous Driving Safety

Korbinian Moller, Rainer Trauth, Johannes Betz

TL;DR

This work tackles occlusion-induced safety challenges in autonomous driving by introducing FRENETIX-Occlusion, a modular module that identifies critical blind spots, models occluded participants as phantom agents, and predicts their movements. It computes a suite of real-time trajectory-safety metrics, including $R(ξ) = max(p(ξ) H(ξ))$ and $BTN = a_{min, req} / a_{veh, max}$, to assess ego-vehicle trajectories and integrate safety considerations into existing planners. The approach is validated on four real-world-inspired CommonRoad scenarios, showing that occlusion-aware planning can reduce risk and prevent collisions with occluded VRUs while maintaining traffic efficiency. The method is released as open-source, enabling further development and potential extensions such as temporal tracking and accelerated implementations.

Abstract

Our work introduces a module for assessing the trajectory safety of autonomous vehicles in dynamic environments marked by high uncertainty. We focus on occluded areas and occluded traffic participants with limited information about surrounding obstacles. To address this problem, we propose a software module that handles blind spots (BS) created by static and dynamic obstacles in urban environments. We identify potential occluded traffic participants, predict their movement, and assess the ego vehicle's trajectory using various criticality metrics. The method offers a straightforward and modular integration into motion planner algorithms. We present critical real-world scenarios to evaluate our module and apply our approach to a publicly available trajectory planning algorithm. Our results demonstrate that safe yet efficient driving with occluded road users can be achieved by incorporating safety assessments into the planning process. The code used in this research is publicly available as open-source software and can be accessed at the following link: https://github.com/TUM-AVS/Frenetix-Occlusion.

Overcoming Blind Spots: Occlusion Considerations for Improved Autonomous Driving Safety

TL;DR

This work tackles occlusion-induced safety challenges in autonomous driving by introducing FRENETIX-Occlusion, a modular module that identifies critical blind spots, models occluded participants as phantom agents, and predicts their movements. It computes a suite of real-time trajectory-safety metrics, including and , to assess ego-vehicle trajectories and integrate safety considerations into existing planners. The approach is validated on four real-world-inspired CommonRoad scenarios, showing that occlusion-aware planning can reduce risk and prevent collisions with occluded VRUs while maintaining traffic efficiency. The method is released as open-source, enabling further development and potential extensions such as temporal tracking and accelerated implementations.

Abstract

Our work introduces a module for assessing the trajectory safety of autonomous vehicles in dynamic environments marked by high uncertainty. We focus on occluded areas and occluded traffic participants with limited information about surrounding obstacles. To address this problem, we propose a software module that handles blind spots (BS) created by static and dynamic obstacles in urban environments. We identify potential occluded traffic participants, predict their movement, and assess the ego vehicle's trajectory using various criticality metrics. The method offers a straightforward and modular integration into motion planner algorithms. We present critical real-world scenarios to evaluate our module and apply our approach to a publicly available trajectory planning algorithm. Our results demonstrate that safe yet efficient driving with occluded road users can be achieved by incorporating safety assessments into the planning process. The code used in this research is publicly available as open-source software and can be accessed at the following link: https://github.com/TUM-AVS/Frenetix-Occlusion.
Paper Structure (13 sections, 12 equations, 8 figures, 6 tables, 1 algorithm)

This paper contains 13 sections, 12 equations, 8 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Required Semantic Information and Framework
  5. Identification of Potential Occluded Traffic Participants
  6. Prediction of Phantom Agent Movement
  7. Occlusion-criticality Measurement
  8. Trajectory Safety Assessment
  9. Results & Analysis
  10. Simulation Setup and Scenarios
  11. Simulation Results
  12. Discussion
  13. Conclusion & Outlook

Figures (8)

  • Figure 1: Exemplary visualization of a critical situation showcases an undetected cyclist emerging from a blind spot at an intersection, presenting a potential crash risk.
  • Figure 2: FRENETIX-Occlusion software module framework with required inputs and provided outputs.
  • Figure 3: Illustration of visible and occluded areas. The visible area $\mathcal{A}_\mathrm{v}$ represents the region within the sensor range that is not obscured by any obstacles. The occluded areas indicate zones occluded by dynamic obstacles $\mathcal{A}_\mathrm{o}$ and boundaries $\mathcal{A}_\mathrm{b}$.
  • Figure 4: Enhanced evaluation funnel for occlusion-aware planning.
  • Figure 5: Scenario 1: A representation of an intersection where an autonomous vehicle navigates visible and occluded zones, highlighting the importance of anticipating PAs' movements for safe trajectory planning.
  • ...and 3 more figures