Autonomous Vehicles Scenario Testing Framework and Model of Computation: On Generation and Coverage

Abstract

Autonomous vehicle (AV) technology started to shift the perception of the transportation systems. However, for AVs to operate at their optimum capabilities, they need to go through a comprehensive testing and verification process. While a large amount of research and funding has been provided for solving this problem, there is still a lack of a systematic method to develop standardized tests that can be used to judge if the decision-making capability functions are within acceptable parameters. To that end, the tests need to cover all possible situations that an AV may run into. This paper focuses on defining the notion of coverage mathematically when using pseudo-randomly generated simulations for testing. The approach defines new equivalence relations between scenes, which are the systems' various states, to achieve this goal. Considering the substantial need for computation, even with the obtained coverage, we also introduce the mathematical definition of a sub-scene and additional strategies, such as expanding the equivalence classes of scenes and combining actors in scenes, to reduce the amount of testing required to certify AVs.

Automatic Generation System for Autonomous Driving Simulation Scenarios Based on PreScan

“…Detailed information about these test scenarios is presented in Table 4. [4] , β weather [2] Yes b ICW HV right turn through intersection, RV straight or turn through intersection…”

“…However, the traditional approach is marred by high testing costs, extended durations, limited scope in realizable scenarios, and low reusability of these scenarios. In contrast, simulation-based scenario testing offers an effective solution, overcoming the limitations of traditional methods while concurrently enhancing the safety of testing procedures [2].…”

Prospects and challenges of Metaverse application in data‐driven intelligent transportation systems

“…[95] reviewed all the publicly available datasets for driving and the virtual testing environments. [96] analyzed the mathematics behind the simulations randomly generated for testing. [97] proposed a framework that utilizes signal temporal logic formulae and evaluates test cases against it for AV systems.…”

“…Despite the amount of research on the topic, there is still a lack of adequate testing methods that can accommodate all possible situations an AV may face [73,[95][96][97][98]. According to [102], it is challenging to test AVs in a very comprehensive and safe manner.…”

Risk analysis of autonomous vehicle test scenarios using a novel analytic hierarchy process method

“…Among various methods to test vehicles, scenario-based testing is the most common approach. In scenario-based testing, critical, complex, and real-world traffic scenarios are applied to verify autonomous functions [1]. To adopt scenario-based testing, there must be a theoretically sound means to evaluate test scenarios to determine if they can realistically represent critical real-world traffic situations.…”