Testing autonomous cars for feature interaction failures using many-objective search

Abstract

Complex systems such as autonomous cars are typically built as a composition of features that are independent units of functionality. Features tend to interact and impact one another's behavior in unknown ways. A challenge is to detect and manage feature interactions, in particular, those that violate system requirements, hence leading to failures. In this paper, we propose a technique to detect feature interaction failures by casting this problem into a search-based test generation problem. We define a set of hybrid test objectives (distance functions) that combine traditional coveragebased heuristics with new heuristics specifically aimed at revealing feature interaction failures. We develop a new search-based test generation algorithm, called FITEST, that is guided by our hybrid test objectives. FITEST extends recently proposed many-objective evolutionary algorithms to reduce the time required to compute fitness values. We evaluate our approach using two versions of an industrial self-driving system. Our results show that our hybrid test objectives are able to identify more than twice as many feature interaction failures as two baseline test objectives used in the software testing literature (i.e., coverage-based and failure-based test objectives). Further, the feedback from domain experts indicates that the detected feature interaction failures represent real faults in their systems that were not previously identified based on analysis of the system features and their requirements.

Testing machine learning based systems: a systematic mapping

“…In case the MLS under test handles scenarios with two or more interacting objects, the input for such a system is a test scenario configuration. For example, in the paper by Abdessalem et al (2018b), the input of the self-driving car simulation is a vector of configurations for each of the objects involved, such as the initial position of the car, the initial position of the pedestrians, the positions of the traffic signs, and the degree of fog.…”

“…In six papers the generation of inputs using a search-based approach aims to detect collision scenarios for autonomous driving systems. Therefore, their fitness functions use metrics such as distance to other static or dynamic objects (Bühler and Wegener 2004;Abdessalem et al 2016Abdessalem et al , 2018b, time to collision (Tuncali and Fainekos 2019;Beglerovic et al 2017;Abdessalem et al 2016), speed of the vehicle (Tuncali and Fainekos 2019;Abdessalem et al 2018b) or level of confidence in the detection of the object in front of the vehicle (Abdessalem et al 2016(Abdessalem et al , 2018a. In contrast, Mullins et al (2018) aim to identify test inputs for an autonomous system that are located in its performance boundaries, i.e., in the regions of the input space where small alterations to the input can cause transitions in the behaviour, resulting in major performance changes.…”

“…Another example of databox access is the work by Groce et al (2014), where the distance between the training data and a test input is used as a metric to address the test selection problem. The remaining 30 cases (43%) require white-box access to the system (Abdessalem et al 2018b;Byun et al 2019;Eniser et al 2019;Kim et al 2019;Ma et al 2018bMa et al , c, d, 2019Murphy et al 2007a, b;Nakajima and Bui 2016;Odena et al 2019;Pei et al 2017;Sekhon and Fleming 2019;Shi et al 2019;Strickland et al 2018;Sun et al 2018a, b;Tian et al 2018;Zhang et al 2018aZhang et al , 2019Li et al 2018;Du et al 2019;Shen et al 2018;Gopinath et al 2018;Guo et al 2018;Xie et al 2019;Cheng et al 2018a;Nakajima 2019), as they need information on the internal state of the trained model. The most prevalent examples in this category are the approaches that need the values of neuron activations to measure some adequacy criteria (Pei et al 2017;Ma et al 2018b) or the weights and biases of the model to apply mutation operators on them (Ma et al 2018d;Shen et al 2018).…”

“…The second most popular category is domain-specific failures defined for autonomous vehicles and driving assistance systems. This set of failures combines several specific instances, the most frequent being deviation from the expected steering angle (five papers (Ma et al 2018c;Patel et al 2018;Pei et al 2017;Tian et al 2018;Zhang et al 2018b)) and number of collisions (10 papers (Abdessalem et al 2018b;Abeysirigoonawardena et al 2019;Beglerovic et al 2017;Bühler and Wegener 2004;Li et al 2016;Majumdar et al 2019;Rubaiyat et al 2018;Tuncali et al 2018;Tuncali and Fainekos 2019;Uesato et al 2019)). The importance of metamorphic and mutation testing for MLS is reflected in the relatively large number of mentions of the associated types of failures: five papers (Xie et al 2018;Murphy et al , 2009Saha and Kanewala 2019;Xie et al 2011) used the violation of metamorphic relationships, whereas three papers (Cheng et al 2018a;Ma et al 2018d;Shen et al 2018) used the notion of mutation killing.…”

“…Unfortunately, only a fraction of authors of the considered studies made their experimental data publicly available (despite we considered the cases when data is partially accessible as "available"). This fraction comprises 13 papers out of 70 (19%) (Abdessalem et al 2018b;Byun et al 2019;Eniser et al 2019;Henriksson et al 2019;Kim et al 2019;Pei et al 2017;Shi et al 2019;Sun et al 2018a, b;Tian et al 2018;Udeshi and Chattopadhyay 2019;Udeshi et al 2018;Zhang et al 2018a). This is a rather negative result, which may hinder the growth of the research in the field.…”