Large-scale Distributed Control

Distributed, embedded, software-based control is a key ingredient of tomorrow's highly versatile systems, which range from microscopic sensors and actuators embedded in our environment to networked home and office appliances to large-scale, distributed systems such as aircraft or entire factories. Developing software for distributed, fault-tolerant, and adaptive control of such systems is enormously complex due to the complexity of the systems to be controlled and the functional demands on the control software (e.g., flexibility, performance, robustness, autonomy). In fact, control software development is the leading cause for project time and cost overruns, and thus can be a major barrier to deploying advanced systems capabilities.

Control means directing a system such that it achieves its objectives within its inherent constraints. Explicit representations of and reasoning about the system and its constraints can dramatically simplify the development as well as increase the capabilities of the control software. The increased power and decreased cost of embedded processors will lead to wide-spread adoption of model-based control, moving from its current usage in slower control problems such as large-scale chemical plant control to faster, complex control problems such as those relevant to advanced, smart systems. The power and flexibility of the model-based approach will lead to more robust control systems developed in a shorter time frame. We believe that constraint programming and model-based reasoning are key enablers of such control software.

In this program, we are working towards the goal of intelligent systems on tomorrow’s reconfigurable modular platforms, platforms which are increasingly equipped with large numbers of embedded sensors, actuators, and computation. With new ideas in embedded reasoning and control, these systems promise exceptionally flexible and robust behavior in dynamic and uncertain environments. Building on model-based approaches using constraint and optimization technologies, our goal is to provide a principled and generic approach to the design and implementation of large-scale distributed control.

This program collaborates with other Smart Matter Integrated Systems projects, including the Modular Robotics project. Also, in the context of DARPA's NEST program, we are developing techniques for distributed adaptive constrained optimization.