This research area covers various fields: from human-centred system design to motion-control; from the identification of dynamic systems to real-time process control, up to the prediction of behaviour in order to implement decision-making systems; from innovative advanced perception methods for autonomous vehicles in unstructured or only partially structured contexts to cooperative estimation and perception strategies for robot networks.
From a methodological point of view, physics-based modelling, hybrid analytics and deep-reinforcement learning techniques are investigated. These techniques are used for Task & Motion planning, for the development of perceptual and cognitive skills of robots and autonomous vehicles, and for the control of robotic systems interacting with the surrounding environment, as well as for the (model-based) control and real-time optimisation of performance and energy consumption of production processes.
Furthermore, these techniques enable the prediction of expected system behaviour, potential anomalies/deviations and resource consumption (electricity, heat, etc.), which can be integrated into event-based and rolling horizon architectures, and also enhance the accuracy of virtual line commissioning solutions through closed-loop simulation.
The research area aims to develop technologies that can be integrated with the most advanced technical solutions, such as distributed/cooperative control solutions or PC-based solutions for motion control and the management of machines for robotic processes, subtractive/additive processes, based on redundant architectures.