Title: Versatile and structurally efficient aerial vehicles assembled from polyhedral rotorcraft modules

Date: Thursday, April 14, 2022

Time: 11:00 AM – 1:00 PM ET

Location: https://bluejeans.com/2450393717

Kévin Garanger

Robotics Ph.D. Student

School of Aerospace Engineering

Georgia Institute of Technology

Committee

Dr. Eric Feron (Advisor) — Computer, Electrical and Mathematical Sciences and Engineering, King Abdullah University of Science and Technology

Dr. John D Berrigan — Materials and Manufacturing Directorate, Air Force Research Laboratory

Dr. Seth A Hutchinson — School of Interactive Computing, Georgia Institute of Technology

Dr. Marco Pavone — Department of Aeronautics and Astronautics, Stanford University

Dr. Julian J Rimoli — School of Aerospace Engineering, Georgia Institute of Technology

Abstract

Autonomous uncrewed air vehicles (UAVs) have become widespread tools for several industries like agriculture, construction, and public safety. They are primarily used to carry payloads such as sensors or cargo. Because of the variety of payloads and the tight performance envelope of electrically powered platforms, a myriad of purpose-built vehicles exist to respond to specific needs, resulting in little flexibility and increase complexity of operations. As a solution to this issue, modular reconfigurable autonomous aerial vehicles that can adapt to diverse payloads have been proposed. Because of the necessity to limit interactions between rotors, prior art has been mostly defined by modules that assemble in a horizontal plane. This rule of assembly inevitably leads to a decrease in stiffness of vehicles as they grow in size and to several related issues. Novel modular rotorcraft designs and assembly schemes based on polyhedral modules and intended to remedy these limitations are explored in this thesis. The proposed rotorcraft modules are inspired by the five platonic solids and in particular, the possibilities offered by tetrahedron and dodecahedron-shaped modules are explored. Both of these modules can form three-dimensional, structurally efficient vehicles showing limited rotor interactions, which are assembled iteratively according to a procedure inspired by fractals.

Structural and dynamical properties of the introduced modular vehicles are studied. The characterization of the polyhedral modules as space frames allows to predict the structural properties of arbitrary vehicles and to compare them. It is shown that overall, three-dimensional configurations are better suited for load-bearing applications. In addition, the potential for a large number of rotors in a vehicle requires novel methods of control allocation. Matrix-based allocation is considered in this work for its low computational cost, and convex optimization programs are formulated to optimize these matrices with respect to metrics such as control authority of power consumption.

Besides being used to study given vehicle configurations, the introduced methods for structural and control allocation analysis can also be integrated into optimization programs over the space of feasible configurations. As a result, it shown that the simultaneous optimization of modular configurations and their control allocation matrices, in consideration of their structural properties as well, can be done with mixed-integer programming. Flight experiments of a prototype show the feasibility of assembling and controlling various configurations from the same modules.