New state estimation/prediction methods will be implemented for both very accurate coupled motion control and large scale system composed of many mobile entities. Novel prediction methods will be developed for systems with real-time constraints. These methods will be able to cope with network delays and bandwidth limitations. Additionally, new decentralised state estimation algorithms, able to cope with different sources of uncertainty and to scale to many mobile entities, will be developed. For estimation purposes, data from monocular and stereo cameras as well as from IMU and GPS sensors will be fused. The project will also develop new, range-only, Simultaneous Localization and Mapping Techniques using the radio signals present in the networked devices, as well as cooperation techniques for intelligent estimation strategies (Active Sensing).
With respect to cooperation, coordination and control techniques, the project will apply in real-time new optimization techniques for scalable and dynamic multi-vehicle route and trajectory planning. The resulting decentralised decision/control algorithms should be able to cope with different sources of uncertainty while, at the same time, scale to a large number of entities. For dynamic routing and trajectory planning, new trajectory optimization techniques will be explored. Furthermore, the above results in distributed estimation/prediction will be used by distributed decision algorithms.
Both estimation and control techniques will be integrated through efficient middleware and communication protocols adapted to satisfy reliability, safety and security for industrial-grade operations. This middleware will integrate both data-centric and network-centric paradigms, and will address the safety challenges by ensuring that synchronisation and temporal constraints in distributed systems for high mobility applications are satisfied. The security of Mobile Ad Hoc Networks for estimation and control will also be considered.