Second Workshop on Quality and Reliability Assessment of Robotic Software Architectures and Components

Today robotic systems are central to many industrial sectors, such as logistics, autonomous warehousing, and healthcare. If on one side ROS is helping roboticists by providing a standardized communication platform for robotic systems, on the other side ROSsystems are getting more and more large and complex, thus making it extremely difficult to ensure their level of quality, e.g., in terms of performance, security, energy efficiency, testability, maintainability. Improving the quality of robotic systems is not a new activity, but in this talk, we tackle it from a different perspective: we look at them from a software architecture perspective. In this talk, I will walk you through a series of experiments we conducted at the Vrije Universiteit Amsterdam targeting the architecture of ROS systems, we will discuss some architectural tactics for ROS systems, and will close with an overview of our open-source tool for automatically executing experiments on robotics software

In the past decade an understanding has arisen in general software engineering that the primary measure of success should be software being successfully used by its intended users. The DevOps movement in particular has highlighted a gap between transferring software systems from the development environment to the operations environment. Recently various researchers and practitioners in robotics have looked at DevOps as a missing part in the way robot applications are created. Robot software faces some specific issues which general software does not have, increasing the complexity of introducing a DevOps way of working. The biggest issue is the physicality of robotics, our machines operate in the real world and thus have real world consequences. To make it easier to test robotic systems in real environments, we have introduced a testing framework called ROSPIT, which allows specifying Physical Integration Tests using a high level XML syntax and graphical tools.

Building robust and reliable robotic software is an inherently challenging feat that requires substantial expertise across a variety of disciplines. Despite that, writing robot software has never been easier thanks to software frameworks such as ROS: Colloquially known as the "Linux of Robotics", ROS drastically reduces the barrier to writing robotics software by facilitating extensive code reuse and collaborative development through its modular design and rich package ecosystem. At its best, ROS allows newcomers to assemble simple, autonomous robots within a matter of hours or days, and domain experts working in a particular area of robotics (e.g., vision) to share their latest algorithms with the community as a reusable software component. In practice, however, truly reusable ROS software components are difficult to build and even trickier to integrate; subtle bugs and misassumptions are easy to introduce and hard to detect. In this talk, I'll discuss how ROS developers build components and systems, the common mistakes that they make, and how lightweight specification and analysis tools can help to identify and prevent those mistakes.

Service robots operate in uncertain and dynamic environments. While being increasingly popular, engineering service robots remains challenging. Especially, evolving them from prototype to deployable product requires effective validation and verification, assuring the robot's correct and safe operation in the target environment. While testing is the most common validation and verification technique used in practice, surprisingly little is known about the actual testing practices and technologies used in the service robotics domain. In this talk I will present observations from an experience report on field testing of an industrial-strength service robot, as it transitions from lab experiments to an operational environment. I will report on challenges and possible solutions, and reflect on their effectiveness in identifying and testing safety-critical scenarios.