Conflict Resolution and Optimization

点击翻译按钮获取中文摘要

The main objectives of Phase 1 aim at developing fundamental algorithmic methods for reconfiguration and reallocation, as well as ensuring that they are useful and suitable for the overall objectives of CCC, by engaging in practical collaboration with the other, more applied projects. This is achieved by considering methods that work at different information and optimization scales (e.g., exact global computation by mathematical programming, heuristic approaches by local methods for gradual improvement), different time scales (e.g., more time-consuming methods applied in an offline setting, faster methods for online scenarios, but also a subdivision into preprocessing/query), as well as different computation scales (e.g., centralized computation vs. game-theoretic mechanism design). These scales are often linked (for example, decentralized computation may correlate with limited information), without being methodologically equivalent (e.g., optimization under limited information arises even in centralized scenarios, giving rise to the area of online algorithms). A general approach in the field of algorithmics is to develop separate methods for scenarios at different scales, i.e., to consider one combination of the described scales, and develop powerful methods that work well under these specific conditions.The main overall objective for Phase 2 is to overcome this limitation and study of multi-scale, integrated and interleaving methods for complex scenarios, in which these different scales (information and optimization, time, computation) are not treated separately, but are employed in an interchanging and parallel fashion. This requires blending and unifying the separate algorithmic methods into multi-scale methods, to be carried out within B1, a close interaction with the conceptual approaches and techniques of distributed contract and algorithm design and employment for complex, structured scenarios in collaboration with other projects, and an integration of the overall methods with the other projects in the context of case studies with the partners from the applied projects. Most notably, these involve studying robot platforms and autonomous vehicles, both of which pose relevant scientific challenges, when organizing resource allocation within one robot or vehicle and when coordinating several platforms. Beyond our algorithmic expertise, we have considerable experience in both application areas, making these very promising targets.