AI for nonlinear control and robotics (T2)

Obiectiv: Deep learning tools for data-driven modeling and end-to-end control of dynamical systems

Research challenges / Novelty / Innovation
• AI and deep neural networks for intelligent nonlinear control
• AI adapted for the control of networked multi-agent systems
• AI for navigation and control of robots and teams of autonomous robots
• AI for state estimation in complex and nonlinear systems

Research results:
• Recurrent neural models adapted for the identification and prediction of nonlinear systems
• Methods for nonlinear optimal control based on recurrent neural models
• Opinion control methods in networks using neural networks adapted to multi-agent structure
• Perception and mapping methods in challenging environments with uncertain sensors
• Robust drone control methods for safety and resilience
• Intelligent planning methods for heterogeneous robot teams
• Real-time state and parameter estimation algorithms for complex and nonlinear systems
• Demonstrator of AI techniques applied to medical rehabilitation

Inovation:
• This topic will extend the interdisciplinary boundaries of AI, control theory, networked systems, and musculoskeletal modeling, leading to innovations that would be impossible to achieve separately in any of these fields.