Towards a Stable 3D Physical Human-Drone Interaction

Key requirements of physical human-drone interactions are that the system is stable, safe, and expressive. The user should be free to interact with the drone in 3D space, and the drone should react appropriately and stably to the physical touch from the user. These requirements are necessary for both single-drone interactions and even more so for the interactions with swarms required to realize a holodeck. The majority of previous physical human-drone interaction systems that have been created use a simple PID controller. Our prior work has shown that these PID controllers are effective at vertical interactions but can quickly become during lateral interactions. However, recent control strategies, such as nonlinear model predictive control (NMPC) and incremental nonlinear dynamic inversion control (INDI) showed improvement in performance in agile flight and handling uncertainties. In this paper, we present the lessons learned from our prior work and discuss implications of these advancements and limitations for physical human-drone interaction. We speculate on how the integration of these advanced control strategies could overcome current limitations, enhancing interaction capabilities. We conclude with suggestions for future research directions, including the exploration of new adaptive methods and their potential integration into human-drone interaction frameworks.