Biologically-inspired computing for coordination and control in large-scale and dynamic systems (BioCo3)
This interdisciplinary project (involving computer science, biology, operations research and logistics expertise) studies decentralized approaches for controlling large-scale, dynamic systems - e.g. for controlling a large number of vehicles that need to accomplish transport tasks for clients (package delivery service). Achieving efficient and effective collective behavior is known to be quite a challenge, esp. if the systems are large in scale and submitted to changing operating conditions.
Complex collective behaviour occurs in many socio-biological systems. Social insects, e.g., are able to find the shortest path to food sources without any central control, or can build enormously complex nests with built-in ventilation systems, or divide the work such that the chances of survival are optimal.
Some socio-biological mechanisms for collective behavior have already
been studied in the context of computing (e.g. stigmergy and digital
pheromones, or in discrete optimization techniques, such as ant colony
optimization, artificial immune systems, genetic algorithms). It becomes
apparent, however, that (1) large-scale and dynamic systems still hold
major challenges, and (2) many socio-biological concepts and techniques
have been unexplored.
The project therefore aims to study advanced concepts and mechanisms, known from social biological systems, and their applicability in decentralized control systems. To guide the research, the project will use concrete application scenarios, based on decentralized control in pickup-and-delivery problems (PDPs).