Socially Responsible CPS Workshop Overview

Cyber-Physical-Human Systems and Social Responsibility Anuradha Annaswamy Department of Mechanical Engineering Active-adaptive Control Laboratory Massachusetts Institute of Technology 1 Socially Responsible CPS, NSF Workshop, November 8, 2022

Cyber Physical Systems Physical: Natural, human- made, obeys physical laws Cyber: Computation, communication, control CYBER-PHYSICAL SYSTEMS Medical Multidisciplin ary, interactive, coordinated Multi-time scale Highly networked New Systems science 2 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPS and Humans Human-in-the-controller Human-in-the-plant Human-machine-symbiosis Human-in-multiagent-loops Interaction with autonomous systems Networked population wide interaction Physiological interaction Figures from Control for Societal-scale Challenges: Roadmap 2030, An IEEE CSS Initiative, Eds: Annaswamy, Johansson, and Pappas, 2022. 3 Socially Responsible CPS, NSF Workshop, November 8, 2022

To be published early 2023 4 Socially Responsible CPS, NSF Workshop, November 8, 2022

Roles of Humans in CPHS A few examples CPHS in Smart grids Humans as empowered prosumers Concept: Distributed Optimization and control Performance: Optimize Grid Services Smart grids + Energy Justice: Modeling, Metrics, Mitigation CPHS in Highway Traffic Humans in the loop (as actuators) Concept: Transactive Control Performance: Congestion alleviation Transportation: Inequities and solutions CPHS in Flight Control Humans as supervisors Shared Decision Making between Pilots & Autopilots Concepts: Capacity for Maneuver & Graceful Command Degradation Humans and Automation: Social Responsibilities 5 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Smart Grids: Energy Justice Consumer is not monolithic Impact of smart grids on low and moderate income (LMI) households not understood The U.S. Department of Energy estimates that households below the federal poverty level spend 18% of their income on energy costs, as compared to 2% for households at 400% of the federal poverty level and above[1]. Net Energy Metering: Low-income households, populations of color, and renters are more likely to be non-participants. Increased adoption of solar PV by high-income homes can increase electricity costs for LMI homes [2] [3]. [1] https://www.energy.gov/eere/slsc/maps/lead-tool [2] Erik Johnson, Ross Beppler, Chris Blackburn, Benjamin Staver, Marilyn Brown, and Daniel Matisoff, Peak Shifting and Cross-Class Subsidization: The Impacts of Solar PV on Changes in Electricity Costs, Energy Policy 106 (2017): 436-444. [3] Eric O Shaughnessy, Galen Barbose, Ryan Wiser, Sydney Forrester, and Na m Darghouth, The Impact of Policies and Business Models on Income Equity in Rooftop Solar Adaption, Nature Energy 6 (2021): 84-91. 6 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Smart Grids: Energy Justice Differential NEM impacts between participating and non-participating customers have equity implications o If you participate in NEM, your electricity bill goes down. Otherwise it does not. o Indirect: Low-income households may not have skills related to renewables (solar vs. coal)[4] Low income households correspond to inefficient buildings with a lot of heat losses. Similar comments apply to renters as well. Lack of informed consent to participate in energy projects Lack of representation in decision making Those without access to broadband or renters with little control over their appliances may find it difficult to adapt to TOU (time-of-use) rates LMI households may struggle to afford smart appliances. Coupling NEM and TOU exacerbates the already limited ability of LMI households to participate in NEM. Emergency load shedding may always affect the same neighborhoods [4] Leon, W., C. Farley, N. Hausman, B. Herbert, N.H. Hammer, B. Paulos, et al. 2019. Solar with justice: strategies for powering up under-resourced communities and growing an inclusive solar market. Clean Energy States Alliance, p. 134. 7 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Smart Grids: Possible solutions Modeling: Use data-driven models to explore disparities in heating energy use intensity[1] Modeling behavioral choices in LMI through a datadriven approach that estimates energy-limiting behavior[2]: o hidden energy poverty (e.g., delaying turning on air conditioning) Electrification planning process needs to include energy justice agent-based approach[3] Microgrids and underserved communities[4] Solar energy certainly offsets pollution and therefore indirectly has a certain value for LMI needs to be quantified [1] Tony G. Reames, Targeting energy justice: Exploring spatial, racial/ethnic and socioeconomic disparities in urban residential heating energy efficiency, Energy Policy 97 (2016): 549-558. [2] Shuchen Cong, Destenie Nock, Yueming (Lucy) Qiu, Bo Xing, Unveiling Hidden Energy Poverty Using the Energy Equity Gap, Nature Communications 13 (2022): 2456. [3] Bethel Tarekegne and Mark Rouleau, An Energy Justice Based Approach for Electrification Planning - An Agent-Based Model, Proceedings of the IEEE Global Humanitarian Technology Conference, 2019. [4] Oluleke Babayomi, Tobi Shomefun, Zhenbin Zhang, Energy Efficiency of Sustainable Renewable Microgrids for Off-Grid Electrification, Proceedings of IEEE PES/IAS PowerAfrica, 2020. 8 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Smart Grids: Possible solutions Metrics and Mitigation: LMI-focused incentives could help reduce solar adoption inequities, while standard incentives risked exacerbating these inequities. Use equity based components in the cost function leading to lower electricity rates Consumer advocacy groups should be involved in the modeling process Coordinate with technologists who should develop quantitative justice metrics 9 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Highway Traffic dynamic toll lanes Sensor: Inductive loop detector Both the traffic volume and traffic speed are measured. Empirically designed toll prices so as to maintain 50mph actual dynamic lane density zero toll lanes Toll structure, MnPass probability of consumer purchase 1 X: 2.525 Y: 0.5126 0.5 0 0 5 10 15 20 price Nonlinear PID desired dynamic lane density $$$ road Transactive Controller driver behavior dynamics * Anuradha M. Annaswamy, Yue Guan, Eric H. Tseng, Zhou Hao, Phan Thao, and Diana Yanakiev. Transactive Control in Smart Cities. Proceedings of the IEEE, Special Issue on Smart Cities, 2017. 10 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Highway Traffic: Inequities & Solutions Dynamic pricing directly impacts on LMI households Appropriate incentive designs are needed Cross-subsidization aspects must be addressed. Consumer/ Rider Highway/Shuttle dynamics Incentive LMI Concerns Consumer Advocacy Groups 11 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS in Flight Control: Social Responsibilities Humans Automation Strategic behavior needs to be accommodated. Compromised human: E and F may be better accomplished by automation Compromised automation: A and B should be addressed by a human Figure adapted from Eraslan, Yildiz, and Annaswamy, Safe Shared Control Between Pilots and Autopilots in the face of Anomalies , In Cyber-Physical-Human Systems: Fundamentals and Applications, IEEE-Wiley, 2023. 12 Socially Responsible CPS, NSF Workshop, November 8, 2022

CPHS Social Responsibility CPHS in Smart grids Energy justice needs to be investigated LMI households: Modeling, Metrics, Mitigation Performance: Optimization framework must include energy justice metrics CPHS in Highway Traffic Transactive control must accommodate the needs of all consumers Cross subsidization aspects need to be addressed Performance: Congestion alleviation CPHS in Flight Control Humans as supervisors Performance: Resilience Adverse actors, both humans and automation, have to be appropriately contended with 13 Socially Responsible CPS, NSF Workshop, November 8, 2022

Thank you! 14 Socially Responsible CPS, NSF Workshop, November 8, 2022