RiskBench: A Scenario-based Benchmark for Risk Identification
Conceitos Básicos
Intelligent driving systems aim to achieve zero-collision mobility through risk identification. RiskBench provides a scenario-based benchmark for evaluating risk identification algorithms.
Resumo
Intelligent driving systems strive for zero-collision mobility by identifying risks from dynamic traffic participants and unexpected events. RiskBench introduces a scenario-based benchmark to assess the performance of ten algorithms in detecting, locating, and anticipating risks. The dataset comprises 6916 scenarios covering various interaction types, actor behaviors, road structures, and weather conditions. The evaluation metrics include risk localization, anticipation, and planning awareness to facilitate decision-making. Existing algorithms face challenges in temporal consistency and robustness in real-world deployment.
RiskBench
Estatísticas
6916 scenarios in the RiskBench dataset.
Evaluation of ten risk identification algorithms.
Metrics include risk localization, anticipation, and planning awareness.
Citações
"RiskBench evaluates an algorithm’s ability to identify risks stemming from dynamic traffic participants and unexpected events based on localization, anticipation, and planning awareness."
"Our aim is to encourage collaborative endeavors in achieving a society with zero collisions."
"We present RiskBench, a benchmark designed for risk identification, with a specific focus on assessing risks stemming from dynamic traffic participants and unexpected events."