Abstract
-
Millions of people use navigation software every day
to commute and travel. In addition, many systems rely upon the
correctness of navigation software to function, ranging from
directions applications to self-driving machinery. Navigation
software is difficult to test because it is hard or very expensive
to evaluate its output. This difficulty is generally known as the
oracle problem, a fundamental challenge in software testing.
In this study, we propose a metamorphic testing strategy to
alleviate the oracle problem in testing navigation software, and
conduct a case study by testing the Google Maps mobile app, its
web service API, and its graphical user interface. The results
show that our strategy is effective with the detection of several
real-life bugs in Google Maps. This study is the first work on
automated testing of navigation software with the detection of
real-life bugs.