Session 1.2: Technological Trends

Session 1.2.A: Engineering Biology

Abstract The field of engineering biology (also referred to as biological engineering or synthetic biology) has emerged over the last 20 years as an exciting domain that makes use of engineering tools to provide better understanding of biological systems, and to design components into biological circuits that carry out useful functions. In this section we describe some of the opportunities for control theory to benefit engineering biology, and highlight unique challenges posed by this application area.

Session 1.2.B: Robotics and Automation

Abstract The first commercially-available automation was likely the automated thermostat, developed in the early twentieth century. By the turn of the twenty first century, computer-programmable robots were widespread in factories worldwide. Although programmable, these complex systems can be very difficult to change in practice. The promise of personalization, customization, and fully flexible automated manufacturing remains unrealized. Service robots, autonomous vehicles, surgical robots and field robots are operating in less structured environments while at the same time taking on more complex and safety-critical tasks. The quest for full autonomy is still ongoing, since current emerging applications call for advanced collaboration, guidance and supervision capabilities that need to account for human presence. Increased levels of resilience, reactivity and reconfigurability are required at the robot control and planning levels and the fusion thereof. In this session, we will address these aspects from both the methodological perspectives as well as the application challenges. 

Session 1.2.C: Electrify Everything

Abstract The recipe for the green energy transition involves substituting fossil energy resources with sustainable power generation, e.g., in terms of PV and wind power. Currently, however, only a small proportion of energy consumption is related to electrical power. Thus, in order to have a fully sustainable energy sector, it is required to electrify all the domains that are currently covered by non-renewable resources. In itself, this comprehensive electrification is a massive undertaking. As this unfolds, however, coupling several infrastructures through electrification should be explored. Such a coupling would offer opportunities to coordinate the use of various infrastructures in order to obtain an overall better use of capacities. This effectively means that the required investments in expanding infrastructure can be substantially reduced through prudent control and coordination approaches. In this session, we will discuss how selected energy-related infrastructures can be electrified individually, and how systems and control can assist in obtaining this. Further, we will discuss how to combine separate infrastructures through appropriate sector coupling mechanisms based on systems and control in order to use available capacity better and in order to reduce the required investments in infrastructure capacity.

Session 1.2.D: AI and Big Data

Regarding large socio-technical systems artificial intelligence (AI), machine leaning (ML) and “big data” science have to be investigated very carefully as their use can be extremely beneficial or extremely harmful. Systems, control and optimization thinking and methods will play a key role as efficient and high-quality inference and decision making are central in this area. We outline several challenges that need to be addressed. The first is the appropriate modeling of human behavior and its implications on decision making. A key difficulty is prediction of human behavior and resulting uncertainties. We have concrete evidence that humans are not causal systems and that their decision-making trades off performance and risk (c.f. Prospect Theory). We have yet to develop quantitative analytics for Prospect Theory and have not investigated its full impact on human-machine collaborative autonomy, although many works have addressed collaborative autonomy. The second, is to integrate ML with knowledge representation and reasoning (KRR), clearly supported by all theories of intelligence and learning for humans and higher-level animals. Such an integrated and concrete view of AI will have tremendous impact on large socio-technical systems (e.g. mixed urban traffic, healthcare delivery, social media). The third has to do with appropriate modeling of large socio-technical systems as distributed, co-evolving multiple graph systems, lacking from the current literature in control systems. We should investigate multiple graph models attempting to capture notions of collaboration, sharing of information and communication of data between the large number of agents involved. The fourth is to address issues of data quality and integrity. The outcomes of AI and ML depend critically on the quality and integrity of the data sources used. We need to develop effective and efficient methods for checking and assessing the quality and integrity of data. The fifth challenge has to do with scalable and inherent measures of trust, security and safety in large socio-technical systems. Nascent efforts exist but much more needs to be investigated and implemented. Such concepts are essential for collaboration and defense against malevolent agents. Finally, there is widely accepted need to investigate these challenges from an integrated system science and systems engineering perspective and address issues of integrated model-based and data-based methods and scalability.