Automate evaluation of regression test results
Ladda ner
Typ
Examensarbete på grundnivå
Program
Mekatronik 180 hp (högskoleingenjör)
Publicerad
2016
Författare
Martinsson, Oscar
Hansson, Fredrik
Modellbyggare
Tidskriftstitel
ISSN
Volymtitel
Utgivare
Sammanfattning
Today, Volvo Cars Corporation uses different sensors to get a 360 degree view of how their cars perceive the surroundings and the environment. As part of the car development, here is a need to test and verify the developed functionality. Currently, SensorFusion department at Volvo Cars Corporation uses automated unit tests (Google Test). However, unit tests are not enough to ensure that the new functionalities do not degrade the product. Regression tests are needed to protect the existing functionality. At the time of this thesis work, Volvo did not know how to validate results of regression tests in an automated environment where one needs a pass / fail test. In this report, we develop a test validation algorithm using regression tests. Our algorithm is validated and evaluates output from tested software by comparing it with simulated data. The simulated data used in this thesis comes from the two sensors from a selected traffic scenarios using object fusion output, which is available at Volvo. In addition to our test validation algorithm, in this thesis we detail our implementation and describe a combination of Matlab files and C ++ code we designed for our work.
Beskrivning
Ämne/nyckelord
Informations- och kommunikationsteknik , Data- och informationsvetenskap , Information & Communication Technology , Computer and Information Science