Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
CPS design with learning-enabled components: a case study
Hartsell C., Mahadevan N., Ramakrishna S., Dubey A., Bapty T., Johnson T., Koutsoukos X., Sztipanovits J., Karsai G.  RSP 2019 (Proceedings of the 30th International Workshop on Rapid System Prototyping, New York, NY, Oct 17-18, 2019)57-63.2019.Type:Proceedings
Date Reviewed: Mar 29 2021

Cyber-physical systems (CPSs) integrate hardware/software components with mechanical/electronic equipment to operate in applications for robotics, avionics, smart grids, and the like. This paper presents a case study for the design of a CPS to control the movement of an autonomous unmanned underwater vehicle (UUV). The UUV tracks “a pipe placed on the seafloor using images from a forward-looking camera.”

Several aspects need to be considered. The first issue is the innate degree of uncertainty due to the complex interactions inside the system and between the system and the environment. The traditional control technologies show their limits and are replaced by CPSs based on learning-enabled components (LECs). Second, the security-critical nature of these applications: they must react correctly even to events that happen only rarely. The design must comply with certification processes, requiring safety assurance arguments backed by substantial evidence. Third, all this complexity requires a framework supporting environment simulation and also testing in the earlier stages, including on the software models. The authors use an assurance-based learning-enabled CPS (ALC) toolchain as the development framework.

Development starts with the modeling of components and messages. ALC utilizes three tools to get the whole architectonic model: the SysML language to define the components as blocks, the robot operating system (ROS) for inter-component communication, and the WebGME infrastructure for instantiating the blocks. Any time the original blocks are modified, ALC updates their instantiations automatically. Each block can have various implementation solutions. All these implementations are evaluated to get the optimum.

The LEC construction follows. ALC allows developers to insert code into blocks. The UUV application uses Python. Data is generated using the Gazebo environment simulator (the authors are currently integrating the SCENIC language for data generation). ALC supports training through artificial neural networks and supervised learning. The goal is to “approximate the ideal mapping function from a set of input variables to a corresponding set of output variables.” The whole process is iterative.

The paper presents the development cycle in detail and evaluates the results obtained for diverse changes of conditions (various architectures for the neural networks, various geometries of the pipeline, and so on). The concluding section identifies possible avenues for future research related to the “formalization and quantitative evaluation of safety case arguments.”

The authors are at Vanderbilt University. The paper level is industry/academia.

Reviewer:  Pierre Radulescu-Banu Review #: CR147227 (2108-0211)
Bookmark and Share
 
General (D.2.0 )
 
Would you recommend this review?
yes
no
Other reviews under "General": Date
Development of distributed software
Shatz S. (ed), Macmillan Publishing Co., Inc., Indianapolis, IN, 1993. Type: Book (9780024096111)
Aug 1 1994
Fundamentals of software engineering
Ghezzi C., Jazayeri M., Mandrioli D., Prentice-Hall, Inc., Upper Saddle River, NJ, 1991. Type: Book (013820432)
Jul 1 1992
Software engineering
Sodhi J., TAB Books, Blue Ridge Summit, PA, 1991. Type: Book (9780830633425)
Feb 1 1992
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy