Detecting What Matters: A Novel Approach for Out-of-Distribution 3D Object Detection in Autonomous Vehicles

2025-06-29

AI Summary

Traditional autonomous driving object detection relies on predefined semantic categories, rendering it incapable of identifying out-of-distribution (OOD) objects—a critical safety vulnerability. To address this, we propose a novel detection paradigm centered on the binary criterion “does this object pose an immediate threat?”, transcending category-dependent recognition. Our method introduces the first end-to-end hazardousness classification framework that jointly integrates 3D object detection, motion trajectory prediction, and dynamic risk assessment. Crucially, it operates without semantic class labels, instead modeling potential hazard solely from geometric position, kinematic state, and interaction intent. This enables robust perception and safe response to previously unseen objects. Experiments demonstrate substantial improvements in hazardous-object detection accuracy under OOD conditions, with significant reductions in both false positives and false negatives. The approach markedly enhances the system’s safety-critical decision-making capability in complex, dynamic traffic environments.

Technology Category

• Computer Vision: Object Detection & Categorization
• Natural Language Processing: Safety and Robustness
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Security and Privacy: Data transparency and provenance

Abstract

Autonomous vehicles (AVs) use object detection models to recognize their surroundings and make driving decisions accordingly. Conventional object detection approaches classify objects into known classes, which limits the AV's ability to detect and appropriately respond to Out-of-Distribution (OOD) objects. This problem is a significant safety concern since the AV may fail to detect objects or misclassify them, which can potentially lead to hazardous situations such as accidents. Consequently, we propose a novel object detection approach that shifts the emphasis from conventional class-based classification to object harmfulness determination. Instead of object detection by their specific class, our method identifies them as either 'harmful' or 'harmless' based on whether they pose a danger to the AV. This is done based on the object position relative to the AV and its trajectory. With this metric, our model can effectively detect previously unseen objects to enable the AV to make safer real-time decisions. Our results demonstrate that the proposed model effectively detects OOD objects, evaluates their harmfulness, and classifies them accordingly, thus enhancing the AV decision-making effectiveness in dynamic environments.

Problem

• Detecting Out-of-Distribution objects in autonomous vehicles
• Shifting from class-based to harmfulness-based object classification
• Improving real-time safety decisions for AVs in dynamic environments

Innovation

• Shifts focus to harmfulness determination
• Uses object position and trajectory
• Detects unseen objects effectively