Test automation through robotic actuation and image analytics

摘要

Internet of Things (IoT) systems include a network of physical devices in communication with each other and, therefore, commonly require cross integration of products from multiple companies. Integration testing can be necessary whenever any central product in the network is updated - hence, the automation of testing is important to save time and resources. In particular, this project focuses on automated testing of security panels for a home security and automation company. Currently, the panels are physically modified so that relays can be used to activate some test scenarios; however, this modification endangers the integrity of the test. Other test scenarios are conducted manually by Quality Engineers. The focus of this paper is on automating testing using robotic actuation and image analytics (IA). The newly designed automated test system (ATS) contains three main subsystems: a test rig that provides fixturing and controls lighting in the test environment, a Cyton Epsilon 300 robotic arm that pushes buttons on a panel and activates specific sensors, and an IA system (IAS) to determine the state of the panel. The IAS takes pictures of the alarm panel in states such as “Armed” or “Disarmed” and determines the panel's state using template matching in OpenCV. Using this feedback, the ATS automatically determines if the panel successfully completed the test. Test results include accuracy testing on the robotic arm, which can mark a piece of paper with a pen 100 times without any noticeable drift and only three misses. Further, template matching was shown to correctly detect six panel states when panel position and lighting were held constant. These findings highlight the importance of a test rig to control position and lighting. Finally, user testing demonstrated a setup time of under 30 minutes, which compares favorably to the day or more required to set up the current testing system.