Automatically Learning Formal Models from Autonomous Driving Software

Yuvaraj Selvaraj,Wolfgang Ahrendt,Martin Fabian,Ghazaleh Panahandeh,Ashfaq Farooqui

doi:10.3390/electronics11040643

Yuvaraj Selvaraj, Wolfgang Ahrendt + Show 3 more

Open Access

https://doi.org/10.3390/electronics11040643

Copy DOI

Journal: Electronics	Publication Date: Feb 18, 2022
Citations: 3	License type: CC BY 4.0

Affiliation: Zenuity (Sweden), Chalmers University of Technology

Abstract

The correctness of autonomous driving software is of utmost importance, as incorrect behavior may have catastrophic consequences. Formal model-based engineering techniques can help guarantee correctness and thereby allow the safe deployment of autonomous vehicles. However, challenges exist for widespread industrial adoption of formal methods. One of these challenges is the model construction problem. Manual construction of formal models is time-consuming, error-prone, and intractable for large systems. Automating model construction would be a big step towards widespread industrial adoption of formal methods for system development, re-engineering, and reverse engineering. This article applies active learning techniques to obtain formal models of an existing (under development) autonomous driving software module implemented in MATLAB. This demonstrates the feasibility of automated learning for automotive industrial use. Additionally, practical challenges in applying automata learning, and possible directions for integrating automata learning into the automotive software development workflow, are discussed.

Highlights

In recent years, the global automotive industry has made significant progress towards the development of autonomous vehicles
This article is an extended version of [34], in which we reported results from a case study to automatically learn a formal model of the Lateral State Manager (LSM), a sub-component of an autonomous driving software programmed in MATLAB
The L∗ algorithm implemented in MIDES ran out of memory during the experiments and could not learn a full model

Summary

Introduction

The global automotive industry has made significant progress towards the development of autonomous vehicles. Such vehicles potentially have several benefits including the reduction of traffic accidents and increased traffic safety [1]. These are highly complex and safety critical systems, for which correct behavior is paramount, as incorrect behavior can have catastrophic consequences. Ensuring safety of autonomous vehicles is a multi-disciplinary challenge, where software design and development processes play a crucial role.

Methods

Results

Discussion

Conclusion