5th Workshop on Design Automation for CPS and IoT (DESTION 2023)
One Day Workshop at IEEE/ACM CPS-IoT WEEK, May 9, 2023, San Antonio, Texas, USA
DESTION 2023 took place on May 9, 2023 as part of the CPS-IoT Week events.
The target audience of DESTION 2023 is researchers and practitioners of Cyber Physical Systems (CPS) design methodologies, machine learning, experts from the tool industry, and end-users from systems companies engaged in CPS and Internet of Things (IoT) development. Over the last few years, there has been transformative progress in AI/ML methods such as learning accurate surrogate models, generative AI, efficient design space exploration, testing and verification, and formal synthesis. This progress coupled with the rapidly growing scale and complexity of CPS and IoT has fueled immense interest in the development of design automation tools. The primary emphasis of the Workshop is on discussing and demonstrating new design tool concepts, methodologies, implementations, and case-studies for design, verification and testing of CPS and IoT.
Cyber-Physical Systems (CPS) such as aircraft, automobiles, industrial robots, medical devices, and Internet-of-Things (IoT) applications, promise significant economic and societal benefits. The design, verification, validation, testing, and operation of such systems present several challenges induced by scale, complexity, uncertainty, and many stringent requirements on safety, performance, security, availability, and many other metrics.
There has been a drastic shift in the manner in which products are designed in the past few decades, from being predominantly mechanical and having independent components to being cyber-physical with highly interacting components. This has resulted in an explosion in the design complexity, leading to very long design cycle times. For several of the complex systems presented above the design process can last years involving several redesign loops. To circumvent this issue, the current state of practice relies on "hot-starting" a new design from a known baseline, which unfortunately limits innovation, preventing a detailed exploration of the design space. The design space, on the other hand, is significantly more complex given the interdependent nature of the multidisciplinary design problem. There have been numerous advances in the area of AI for Design Automation methods that have been shown to help in the design of these complex systems, as well as in their autonomous operations. These methods range from natural language processing for requirements engineering, physics-informed models to accelerate simulations, Bayesian methods for uncertainty quantification, probabilistic programming methods to represent designs as a handful of examples. On the other hand, as AI is integrated into a diverse variety of systems such as autonomous vehicles, energy grids, health care, IoTs, and social network platforms, the challenge of design and verification of AI-enabled systems has become extremely important. This has led to new Design Automation for AI methods of interest including network architecture exploration techniques, AI testing and verification methods, and simulation tools.
DESTION provides a premier forum for researchers and engineers from academia, industry, and government to present and discuss challenges, promising solutions, and applications in design automation for CPS and IoT. DESTION 2023 has a broad scope covering techniques and tools for modeling, simulation, synthesis, validation, and verification of CPS and IoT, with a focus on "AI for Design Automation" and "Design Automation for AI", and their applications in a variety of domains, such as automotive and transportation systems, avionics, robotics, building architectures, grid, and medical devices.
Workshop Schedule (May 9, 9:00AM-5:30PM)
|9:10AM-10:00AM||Invited Talk: Designing for designers. Ankur Mehta (UCLA)|
|10:30AM-10:45AM||Hamiltorch: A PyTorch-based library for Hamiltonian Monte Carlo. Adam Cobb|
|10:45AM-11:00AM||Software Introspection for Signaling Social-Cyber Operations. Huascar Sanchez and Briland Hitaj|
|11:00AM-12:00PM||Invited Talk: Teaching AI Co-Designers the Right Lessons: An Urban Air Mobility Case Study. Sydney Whittington (SwRI)|
|01:30PM-01:55PM||Reusable Network Simulation for CPS Co-Simulations. Himanshu Neema, Harmon Nine and Thomas Roth|
|01:55PM-02:20PM||Surrogate Modeling using Physics-guided Learning. Ali Ozdagli, Peter Volgyesi and Xenofon Koutsoukos|
|02:20PM-02:45PM||Constrained Bayesian optimization for Automatic Underwater vehicle hull design. Harsh Vardhan, Peter Volgyesi, Will Hedgecock and Janos Sztipanovits|
|02:45PM-03:00PM||MegaFlow2D: A Parametric Dataset for Machine Learning Super-resolution in Computational Fluid Dynamics Simulations. Wenzhuo Xu, Noelia Grande Gutiérrez and Christopher McComb|
|03:30PM-03:55PM||AIMED: AI-Mediated Exploration of Design: An Experience Report. Sanjai Narain, Dana Chee, Pranav Iyer, Emily Mak, Ricardo Valdez, Manli Zhu, Niraj Jha, Jaime Fisac, Kai-Chieh Hsu, Prerit Terway, Kishore Pochiraju, Brendan Englot, Emil Pitz, Sean Rooney and Yewei Huang|
|03:55PM-04:20PM||Middleware for a Heterogeneous CAV Fleet. Matthew Nice, Matthew Bunting, Daniel Work and Jonathan Sprinkle|
|04:20PM-04:45PM||Tackling Simulation Inconsistencies in the Robot Design Process by Selective Empirical Evaluation. Anwesha Chattoraj, Eric Vin, Yusuke Tanaka, Jillian Naldrien Pantig, Daniel J. Fremont and Ankur Mehta|
|04:45PM-05:10PM||Symbiotic CPS Design-Space Exploration through Iterated Optimization. Sheng-Jung Yu, Inigo Incer, Valmik Prabhu, Anwesha Chattoraj, Eric Vin, Daniel Fremont, Ankur Mehta, Alberto Sangiovanni-Vincentelli, Shankar Sastry and Sanjit Seshia|
|05:10PM||Open Discussion and Closing Remarks|
Abhishek Dubey (Vanderbilt University, USA
Alessio Lomuscio (Imperial College London)