Opening a Hinged Door with an Aerial Manipulator using Model Predictive Control

1 minute read

From aerial pick-and-place to aerial transportation, aerial manipulation has been extensively stud-ied in recent years thanks to its mobility and dexterity, each of which is a merit of an aerial vehicleand a robotic arm. However, to fully harness the concept of aerial manipulation, more complextasks including interaction with movable structures should also be considered. Among varioustypes of movable structures, this paper presents a multirotor-based aerial manipulator opening adaily-life moving structure, a hinged door. Two additional issues that would arise in interactionwith a movable structure are addressed: 1) a constrained motion of the structure, and 2) collisionavoidance with a moving structure. To handle these issues, we formulate a model predictive control(MPC) problem with a system dynamics constraint and state constraints for collision avoidance.A coupled system dynamics of a multirotor-based aerial manipulator and a hinged door is derivedand later simplified for faster computation in MPC. State constraints for collision avoidance withitself, a door, and a doorframe are generated. By implementing a constrained version of differentialdynamic programming (DDP), we can generate reference trajectories through MPC in real-time.The proposed method is validated through simulation results and experiments with a real-likehinged door in which a disturbance observer (DOB) based robust motion controller is employed.

Bibtex

@phdthesis{이동재2020opening,
  title={Opening a Hinged Door with an Aerial Manipulator using Model Predictive Control},
  author={이동재},
  year={2020},
  school={서울대학교 대학원}
}

Categories:

Updated: