Derek James Bennet

Verifiable control of a swarm of unmanned aerial vehicles

Abstract This article considers the distributed control of a swarm of unmanned aerial vehicles (UAVs) investigating autonomous pattern formation and reconfigurability. A behaviour-based approach to formation control is considered with a velocity field control algorithm developed through bifurcating potential fields. This new approach extends previous research into pattern formation using potential field theory by considering the use of bifurcation theory as a means of reconfiguring a swarm pattern through a free parameter change. The advantage of this kind of system is that it is extremely robust to individual failures, is scalable, and also flexible. The potential field consists of a steering and repulsive term with the bifurcation of the steering potential resulting in a change of the swarm pattern. The repulsive potential ensures collision avoidance and an equally spaced final formation. The stability of the system is demonstrated to ensure that desired behaviours always occur, assuming that at large separation distances the repulsive potential can be neglected through a scale separation that exists between the steering and repulsive potential. The control laws developed are applied to a formation of ten UAVs using a velocity field tracking approach, where it is shown numerically that desired patterns can be formed safely ensuring collision avoidance.

Autonomous three-dimensional formation flight for a swarm of unmanned aerial vehicles

This paper investigates the development of a new guidance algorithm for a formation of unmanned aerial vehicles. Using the new approach of bifurcating potential fields, it is shown that a formation of unmanned aerial vehicles can be successfully controlled such that verifiable autonomous patterns are achieved, with a simple parameter switch allowing for transitions between patterns. The key contribution that this paper presents is in the development of a new bounded bifurcating potential field that avoids saturating the vehicle actuators, which is essential for real or safety-critical applications. To demonstrate this, a guidance and control method is developed, based on a six-degreeof-freedom linearized aircraft model, showing that, in simulation, three-dimensional formation flight for a swarm of unmanned aerial vehicles can be achieved.

Distributed control of multi-robot systems using bifurcating potential fields

The distributed control of multi-robot systems has been shown to have advantages over that of conventional single-robot systems. These include scalability, flexibility and robustness to failures. This paper considers pattern formation and reconfigurability in a multi-robot system using a new control algorithm developed through bifurcating potential fields. It is shown how various patterns can be achieved autonomously through a simple free parameter change, with the stability of the system proven to ensure that desired behaviours always occur.

Three -Dimensional Formation Flying Using Bifurcating Potential Fields

This paper describes the design of a three-dimensional formation flying guidance and control algorithm for a swarm of autonomous Unmanned Aerial Vehicles (UAVs), using the new approach of bifurcating artificial potential fields. We consider a decentralized control methodology that can create verifiable swarming patterns, which guarantee obstacle and vehicle collision avoidance. Based on a steering and repulsive potential field the algorithm supports flight that can transition between different formation patterns by way of a simple parameter change. The algorithm is applied to linear longitudinal and lateral models of a UAV. An experimental system to demonstrate formation flying is also developed to verify the validity of the proposed control system.

Enabling and Exploiting Self-Similar Central Symmetry Formations

In this work a formation flying-based architecture is presented within the context of a distributed antenna array. An artificial potential function method is used to control the formation, whereby deviation from an all-to-all interaction scheme and swarm shaping are enabled through a self-similar connection network. Introduction of an asymmetric term in the potential function formulation results in the emergence of structures with a central symmetry. The connection network then groups these identical structures through a hierarchical scheme. This produces a fractal shape that is considered for the first time as a distributed antenna array exploiting the recursive arrangement of its elements to augment performance. A 5-element Purina fractal is used as the base formation, which is then replicated a number of times increasing the antenna array aperture and resulting in a highly directional beam from a relatively low number of elements. Justifications are provided in support of the claimed benefits for distributed antenna arrays exploiting fractal geometries. The formation deployment is simulated in Earth orbit together with analytical proofs completing the arguments aimed to demonstrate feasibility of the concept and the advantages provided by grouping antenna elements into coherent structures.

An analysis of dissipation functions in swarming systems

Abstract Swarms of multiple, autonomous mobile agents have been shown to have advantages over single agent systems such as scalability, robustness and flexibility. This papers considers swarm pattern control using a generic artificial potential field and a range of dissipation control terms. An investigation of a number of dissipation terms to induce different swarm behaviours is undertaken. In addition, a novel dissipation control term is introduced based on time-delay feedback control. It is shown that a delayed dissipation term can induce vortex formations without knowledge of relative velocities. Finally, a stability analysis is undertaken that verifies swarm behaviour in a subset of these cases.

Enhancing self-similar patterns by asymmetric artificial potential functions in partially connected swarms

The control of mobile robotic agents is required to be highly reliable. Artificial potential function (APF) methods have previously been assessed in the literature for providing stable and verifiable control, whilst maintaining a high degree of nonlinearity. Further, these methods can, in theory, be characterised by a full analytic treatment. Many examples are available in the literature of the employment of these methods for controlling large ensembles of agents that evolve into minimum energy configurations corresponding in many cases to regular lattices [1-2]. Although regular lattices can present naturally centric symmetry and self-similarity characteristics, more complex formations can also be achieved by several other means. In [3] the equilibrium configuration undergoes bifurcation by changing a parameter belonging to the part of artificial potential that couples the agents to the reference frame. In this work it is shown how the formation shape produced can be controlled in two further ways, resulting in more articulated patterns. Specifically the control applied is to alter the symmetry of interactions amongst agents, and/or by selectively rewiring interagent connections. In the first case, the network of connections remains the same, and may be fully connected.