Bootstrapping MDE Development from ROS Manual Code - Part 2: Model Generation

Abstract

In principle, Model-Driven Engineering (MDE) addresses central aspects of robotics software development. Domain experts could leverage the expressiveness of models; implementation details over different hardware could be handled by automatic code generation. In practice, most evidence points to manual code development as the norm, despite several MDE efforts in robotics. Possible reasons for this disconnect are the wide ranges of applications and target platforms making allencompassing MDE IDEs hard to develop and maintain, with developers reverting to writing code manually. Acknowledging this, and given the opportunity to leverage a large corpus of opensource software widely adopted by the robotics community, we pursue modeling as a complement, rather than an alternative, to manually written code. Our previous work introduced metamodels to describe components, their interactions, and their resulting composition, as inspired by, but not limited to, the de-facto standard Robot Operating System (ROS). In this paper we put such metamodels into use through two contributions [1]. First, we automate the generation of models from manually written artifacts through extraction from source code and runtime system monitoring. Second, we make available an easy-to-use web infrastructure to perform the extraction, together with a growing database of models so generated. Our aim with this tooling, publicly available both as-a-service and as source code, is to lower the MDE barrier for practitioners and leverage models to 1) improve the understanding of manually written code; 2) perform correctness checks; and 3) systematize the definition and adoption of best practices through large-scale generation of models from existing code. A comprehensive example is provided as a walk-through for robotics software practitioners.

Applying MDE to ROS Systems: A Comparative Analysis

“…The same benefits are claimed by ROS Model [12,11], a DSL allowing to design robot applications. Its main platform target is ROS, but it allows the possibility to switch target later.…”

Bootstrapping MDE development from ROS manual code: Part 2—Model generation and leveraging models at runtime

“…We have demonstrated in previous publications [33,34] the validity of our metamodels to design ROS robot applications, and we have consolidated our concept with all the complementary tools (see Sect. 3.2) that shows to the ROS community the benefits of the use of MDE technologies.…”

“…In [33], we analyzed the models auto-generated using both our methods (the static analysis of the code and the introspection at runtime of the application) on real robot systems. Considering the known limitations of both methods based on their design, the comparison can be shortly summarized by:…”

“…After an analysis of the results and the use of the approach to analyze real use cases, we identified not only the shortcomings of our approach, but we also found an interesting and promising new application domain: the diagnosis of robotics systems at runtime. For this paper, in comparison with our previous publication [33], we introduce, as novel contribution, new features to mitigate partly or completely the limitations found for both previous efforts, i.e., (1) the techniques to extract models and (2) the metamodels definition, and also we show our first contributions to leverage our efforts to monitor large ROS systems during their execution.…”

“…6, we compare the pros and contras of both extraction methods and present one of the main novel contributions of this paper, a tool to auto-combine the results obtained by both methods. This tool is based on lessons learned since our first experiments with model extractors [33]. Section 7 presents our second main new contribution: An adaptation of our metamodels to cover additional relevant runtime information required for performance analysis at system level and first experiments about how our models and infrastructure can be used to diagnose the performance of a running system during its execution.…”