Control of light vehicle-manipulator systems

ongoing

The advent of light and freely moving autonomous robots that can perform complex and versatile operations in challenging environments, involving highly precise and forceful physical interactions, would represent a scientific breakthrough. In such robots, kinematics and dynamics will be intimately connected. Although solutions for specific use cases do exist, a control-theoretical framework, capable of dealing with this tight coupling in a wide range of contexts, is lacking. In CRÈME, I will establish the foundations for such a general framework for autonomous control of systems with strong coupling of kinematics and dynamics. In three research lines, I will build the framework while gradually increasing the level of complexity. In parallel, I will evaluate and demonstrate the utility of this framework for motion planning and control for energy-efficient propulsion, hovering, forceful interaction with objects, and energy harvesting in articulated autonomous underwater vehicles. The rationale for using underwater vehicles as a demonstration arena is partly due to my familiarity with the field, but primarily because the control challenges are particularly demanding. Thus, this will be the ‘litmus test’ for the utility of the framework in a whole range of other contexts. Over three decades, I have contributed with fundamental research to the control theory of underactuated marine vehicles and snake robot locomotion. This has brought me thorough understanding of both the practical challenges of robots in marine environments and the strengths and limitations of current control theory frameworks. More importantly, it strengthened my conviction that the field needs a leap forward as described above. Hence, this project represents an exciting and ambitious endeavour with great potential impact. Despite being hence of a high-risk, high-gain nature, this project is structured such that it will provide pioneering control theoretical contributions for this new class of robots.