Whole-Body Control and Teleoperation of a Suspended Aerial Manipulator
Andre Coelho is a PhD student in the department Robotics and Mechatronics. (Co)Promotors are prof.dr.ir. A. Franchi from the faculty of Electrical Engineering, Mathematics and Computer Science and dr.ing. K. Kondak and prof.dr.ing. C. Ott from the German Aerospace Center.
In recent years, the field of aerial manipulation has witnessed significant advancements in all its subareas. Among the diverse designs, suspended aerial manipulators, like the DLR Suspended Aerial Manipulator (SAM), emerged as promising solutions due to their inherent advantages in terms of energy efficiency, safety, payload capacity, and redundancy.
By harnessing the versatility of complex robots while circumventing the challenges of fully autonomous operation, telemanipulation has historically proven to enable the execution of complex tasks. However, the time-delayed teleoperation of redundant aerial robots like the SAM poses important challenges not yet tackled in the literature. These challenges encompass the system's redundancy, the plurality of actuators, the absence of full-pose measurements, and the limited thrust of the propellers.
In that scope, this thesis aims at achieving the following goals: (1) Adapting whole-body control approaches developed for fixed-base robots to aerial robots; (2) enabling time-delayed bilateral teleoperation of redundant aerial robots, allowing the operator to exploit the system redundancy and command multiple tasks. (3) Estimating the full pose of nonlinear systems in the presence of Pfaffian constraints. (4) Addressing limited propeller actuation and developing energy-efficient dynamic maneuver control techniques to reach the full workspace of suspended aerial robots.
Initially, we present a control system that compensates for the motion of the platform's center of mass and allows for its camera view to be commanded as a secondary task. Subsequently, a passivity-based framework for whole-body bilateral teleoperation of redundant robots is introduced. To estimate the pose of the system, an observer for multi-body systems subject to Pfaffian constraints is presented. Moreover, a Model Predictive Control (MPC) framework, called EigenMPC, capable of finding sustained nonlinear oscillations is introduced.
The proposed teleoperation framework is the first to allow for multi-task and multi-device bilateral teleoperation in the presence of time delays. Additionally, the EigenMPC framework enables limited-actuation systems to exploit their full workspace. The methods presented here are validated in both indoor and outdoor experiments. The results demonstrate an integrated system using information from a variety of sensors and cameras, receiving commands from a remote operator, and using three distinct actuation types to perform manipulation tasks.
