Styrning och Driftsättning av Autonoma Tjänsterobotar

This thesis investigates the control, modeling, and localization of unmanned service robots to enable their autonomous operation in real-world applications. Recent developments in electronics and computational capability have enabled the inspection and maintenance via aerial and ground robotic platforms, thus creating novel research opportunities in the area of control frameworks of such robotics setups. This thesis aims to provide the foundation for deploying novel service robots, in the form of climbing robots and morphing aerial vehicles in the robotics community. Towards this envisioned aim, this thesis establishes the following theoretical and practical contributions: 1) A framework for the autonomous navigation of a vortex actuated climbing robot on both flat and curved surfaces in all orientations for user-predefined missions, leveraging a simple point-to-point tracking approach. In this case, the adhesion requirements are modeled in this framework based on the robot’s payload, surface characteristics, and orientation. At the same time, an adaptive control scheme is developed for the identified vortex actuator characteristics to regulate the adhesion requirements. 2) A novel geometry aware Nonlinear Model Predictive Control (NMPC) scheme that allows foldable quadrotors traversing narrow entrances and tight openings. The developed scheme relies on an augmented objective function that considers the size of an opening and optimizes in flight the configuration of Micro Aerial Vehicle (MAV), while considering the kinematic models of the MAV’s arms. Furthermore, a switching Model Predictive Control (MPC) is defined for the linearized Newton-Euler kinematic equations as a Linear Parameter-Varying (LPV) system that can regulate the attitude of such foldable quadrotors. 3) A range-aided flexible relative pose estimation scheme for multiple robots, where the developed framework estimates the position or pose between two robots without requiring global position information of the localized robot. The localization scheme relies on Ultra WideBand (UWB) transceivers and the pose of the localizer robot that can identify the position of objects with installed UWB sensors with respect to its frame of reference. 4) An external force estimator and disturbance rejection scheme, based on the Nonlinear Moving Horizon Estimation (NMHE) approach that deals with the extreme dynamic effects of the reconfiguration, wind gusts, and other exogenous disturbances. 5) Finally, a multi-stage NMPC scheme to deal with unknown but bounded communication delays between a MAV and the edge. This framework considers a tree of discrete equivalent systems at different sampling times to address the random delays in the communication link between the MAV and the edge, which otherwise would result in an uncertain control actions period. The efficacy of the proposed frameworks is evaluated under multiple experimental and simulation studies