Uncertainty-Aware Vision-Based Risk Object Identification Via Conformal Risk Tube Prediction
KAI-YU FU, Yi-Ting Chen
AI summary
Problem
Existing Vision-ROI methods rely on deterministic predictions that ignore uncertainty, causing premature or delayed risk detection and fragmented outputs in ambiguous, multi-risk driving scenarios.
Approach
The framework represents risk as a spatiotemporal tube and applies category-aware conformal prediction with a feature-alignment loss to calibrate risk scores and guarantee predictive coverage across diverse hazard types.
Key results
- Introduces Conformal Risk Tube Prediction for reliable spatiotemporal uncertainty modeling
- Constructs the Multiple Coexisting Risks dataset for multi-risk scenario evaluation
- Achieves higher calibrated risk coverage and tighter temporal alignment than baselines
- Reduces nuisance braking alerts and improves downstream safety-critical performance
Why it matters
Provides intelligent driving systems with statistically guaranteed, uncertainty-aware risk assessments to prevent safety-critical failures in complex traffic.
Abstract
We study object importance-based vision risk object identification (Vision-ROI), a key capability for hazard detection in intelligent driving systems. Existing approaches make deterministic decisions and ignore uncertainty, which could lead to safety-critical failures. Specifically, in ambiguous scenarios, fixed decision thresholds may cause premature or delayed risk detection and temporally unstable predictions, especially in complex scenes with multiple interacting risks. Despite these challenges, current methods lack a principled framework to model risk uncertainty jointly across space and time. We propose Conformal Risk Tube Prediction, a unified formulation that captures spatiotemporal risk uncertainty, provides coverage guarantees for true risks, and produces calibrated risk scores with uncertainty estimates. To conduct a systematic evaluation, we present a new dataset and metrics probing diverse scenario configurations with multi-risk coupling effects, which are not supported by existing datasets. We systematically analyze factors affecting uncertainty estimation, including scenario variations, per-risk category behavior, and perception error propagation. Our method delivers substantial improvements over prior approaches, enhancing vision-ROI robustness and downstream performance, such as reducing nuisance braking alerts. For more qualitative results, please visit our project webpage: https://hcis-lab.github.io/CRTP/