Sanjai Narain
ABSTRACT
The goal of DARPA's Symbiotic Design of Cyber Physical Systems (SDCPS) program is to develop tools for “correct-by-synthesis” design of cyber physical systems (CPS) and reduce the time from concept to deployment from years to months. Achieving this goal poses several hard challenges. Design spaces are high-dimensional cross-products of discrete and continuous spaces. It can take minutes to hours to evaluate the performance of a design. The human designer's intent is often not concretely articulated. Sometimes designs are not created from scratch but rather by completing or repairing existing ones. This paper outlines how the AIMED system addresses these challenges. AIMED consists of three core technologies. The first is “deformable connector” that eliminates an important type of discreteness from design spaces. Thus, not only is the design space vastly simplified, efficient optimization engines for purely continuous spaces can be used in the search for a design. The second core technology is Inverse Specification, based on inverse reinforcement learning that infers human intent by asking the human a small number of simple preference questions. The third core technology is Gaussian Mixture Models that allows completion and repair of designs and finds not just one but a diversity of solutions. AIMED is illustrated in the context of Unmanned Airborne Vehicles (UAVs) although it was also applied to the design of Unmanned Underwater Vehicles (UUVs). AIMED was used to automatically discover high-scoring, novel UAVs, unencumbered by biases of planarity and symmetry: a UAV with non-coplanar propellers and another with asymmetric wings. We expect our experience will apply to design of other CPS.
- Biyik, Erdem., Huynh, N., Kochenderfer, M. J., and Sadigh, D., Active preference-based Gaussian process regression for reward learning, Proceedings of Robotics: Science and Systems (RSS), Jul. 2020.Google Scholar
Cross Ref
- Christiano, Paul, Leike, J., Brown, T. B., Martic, M., Legg, S., and Amodei, D., Deep reinforcement learning from human preferences, arXiv: 1706.03741 [stat.ML], 2017.Google Scholar
- Deb, Kalyanmoy, Pratap, A., Agarwal, S., and Meyarivan T., A fast and elitist multiobjective genetic algorithm: NSGA-II, IEEE Transactions on Evolutionary Computation, vol. 6, no. 2, pp. 182-197, 2002.Google ScholarDigital Library
- Ahmed, Mohamed, and Prince, S., “Bayesian optimization,” Borealis AI. https:/www.borealisai.com/en/blog/tutorial-8- bayesian-optimization/Google Scholar
- Nandakumar, Ganeshram, Ranganathan, T., Arjun, B. J. and Thondiyath, A. design and analysis of a novel quadrotor system - VOOPS, IEEE International Conference on Robotics and Automation (ICRA), 2015.Google ScholarCross Ref
- Narain, Sanjai, Mak, E., Chee, D., Englot, B., Pochiraju, K., Jha, N. K. and Narayan, K., 2021. Fast design space exploration of nonlinear systems: Part I. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 41(9), pp. 2970-2983. https://arxiv.org/pdf/2104.01747Google Scholar
- Ryseck, Peter. Extending Quadcopter Flight Time—By Adding Wings. shorturl.at/eswCKGoogle Scholar
- Narain, Sanjai, and Arkoudas, K., Optimal Network Configuration Repair. US Patent US 8,725,902 B2. https://patents.google.com/patent/US8725902B2/en. 2014.Google Scholar
- Terway, Prerit, Hamidouche, K. and Jha, N. K., 2021. Fast design space exploration of nonlinear systems: Part II. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 41(9), pp. 2984-2999. https://arxiv.org/pdf/2104.02464Google Scholar
- Terway, Prerit, and Jha, N. K. REPAIRS: Gaussian mixture model based completion and optimization of partially-specified systems. Under review in ACM Transactions on Embedded Computing Systems.Google Scholar
- Bapty, Ted, Whittington, S., Walker, J., Hite, J., Swenson, B., Owens, K., Eisele, F., Scott J., and Owens, R., Design oracle for AI-based CPS design. Proceedings of Design Automation for CPS and IoT (DESTION 2022).Google ScholarCross Ref
- Z3: An Efficient SMT Solver. https://www.microsoft.com/en-us/research/project/z3-3/Google Scholar
- Ziebart, Brian, Maas, A., Bagnell, J. A., and Dey, A. K., Maximum entropy inverse reinforcement learning, Proceedings of the 23rd National Conference on Artificial Intelligence (AAAI), Chicago, Illinois, 2008.Google Scholar
Recommendations
Making Sense of Body and Space through Full-Body Interaction Design: A Case Study
IDC '16: Proceedings of the The 15th International Conference on Interaction Design and ChildrenResearch based on the embodied cognition framework proposes the use of physicality and spatiality to promote learning. Nevertheless, methods to include children in the design of Full-Body Interaction Learning Environments often neglect to properly ...
Applying Storycraft to Facilitate an Experience-Centric Conceptual Design Process
Design, User Experience, and Usability. Interaction DesignAbstractThe design profession is shifting from designing objects towards designing for experiences, and the main premise of this paper is that designers need strategic guidance in bringing the emotional, contextual, and temporal aspects of experiences ...
Participatory development of user experience design guidelines for a B2B company
AcademicMindtrek '16: Proceedings of the 20th International Academic Mindtrek ConferenceAs business success is increasingly dependent on an organization's ability to provide a pleasant user experiences (UX) for its products, companies need to find ways to harness every employee to think about UX in their daily work. To support this goal, ...
Comments