Most of the assessment of risk from occlusion in autonomous vehicles (AV) has been so far focused on static occlusion, i.e., occlusions caused by trees, buildings, parked cars, etc.
On the other hand, situations of dynamic occlusion (occlusion caused by another vehicle in traffic) have unique challenges and can appear unexpectedly at any moment in traffic. Therefore, a recent study presents a novel safety validation framework for strategic planners in AV.
The researchers used the theory of hypergames to develop a novel multi-agent measure of situational risk. The hypergames theory expands standard game theory by proposing a hierarchical framework. At higher levels, agents have a higher awareness about other agents’ views of the game that may not match their own.
The experimental results show that the proposed validation method achieves a 4000% gain in generating occlusion causing crashes compared to naturalistic data only.
A particular challenge for both autonomous and human driving is dealing with risk associated with dynamic occlusion, i.e., occlusion caused by other vehicles in traffic. Based on the theory of hypergames, we develop a novel multi-agent dynamic occlusion risk (DOR) measure for assessing situational risk in dynamic occlusion scenarios. Furthermore, we present a white-box, scenario-based, accelerated safety validation framework for assessing safety of strategic planners in AV. Based on evaluation over a large naturalistic database, our proposed validation method achieves a 4000% speedup compared to direct validation on naturalistic data, a more diverse coverage, and ability to generalize beyond the dataset and generate commonly observed dynamic occlusion crashes in traffic in an automated manner.
Link to the article: Kahn, M., Sarkar, A., and Czarnecki, K., “I Know You Can’t See Me: Dynamic Occlusion-Aware Safety Validation of Strategic Planners for Autonomous Vehicles Using Hypergames”, 2021 https://arxiv.org/abs/2109.09807