H2AMI: Intuitive Human to Aerial Manipulator Interface

Aerial manipulators are unmanned aerial vehicles (UAVs) coupled with robotic manipulators (RMs). Their ability to interact with the environment and perform service and maintenance tasks in hard-to-reach places makes them interesting for a wide range of applications and industries. Human pose estimation relates to the problem of estimating human body pose. In this paper, we present an intuitive aerial manipulator human interface which enables the control of the aerial manipulator based on the operator's movements. The operator's motions are captured with an affordable depth camera and pose estimation is performed with a neural network. We have developed a system that enables intuitive decoupled control of the unmanned aerial vehicle and robotic manipulator. We compared two neural network architectures for the human pose estimation as a part of the H2AMI. System performance was verified in the simulation environment.