Modeling and Control in Mobile Robotics: Nonholonomic Constraints, Predictive Control, and Collaboration

Robotic systems are transitioning from rigid, isolated automation cells toward mobile, distributed, and physically interacting systems operating in dynamic environments. Wheeled mobile robots are a key enabling technology in this transition, supported by major advances in perception and localization. However, progress in control and coordination has not yet been fully realized. Mobile robotics still relies predominantly on classical control techniques, while multi-robot systems remain limited in deliberately coordinating physical interactions. Against this background, this thesis develops tailored optimization-based control methods for wheeled mobile robots and advances physically coupled collaboration in robot teams. By explicitly exploiting the geometric structure induced by nonholonomic kinematics, it introduces model predictive control formulations for high-precision setpoint control of individual robots and formations, broadly applicable across different platforms. It further investigates data-driven models of real-world system behavior and integrates them into the framework in a geometry-consistent fashion. Extending from motion control to physical interaction, the thesis advances physically coupled collaboration through force-based strategies for cooperative object transportation. By employing optimization-based methods for both control and organization, the approach is generic and scalable for multi-robot systems. Beyond the applications studied, the geometry-conforming design principles are directly relevant to motion planning and control of automotive vehicles. Moreover, organizing multi-robot systems through explicit interaction-force control opens new perspectives in distributed robotics, including flexible-object transportation and other manipulation tasks.