Get started
Work

Extended Swarming

Year: 2026
Type: dissertation
Abstract: Robot swarms are composed of locally interacting units that operate without central control, resulting in high scalability, robustness against failures, and flexibility. This makes them particularly suitable for scenarios in which a large, challenging, and dynamic environment must be covered simultaneously, such as search and rescue operations. A key challenge in swarm robotics is enabling a single human operator to control and understand a large swarm without becoming overwhelmed by its complexity, while preserving the swarm's decentralised benefits. Most existing work in Human–Swarm Interaction (HSI) has focused either on the technical aspects of swarm control or on human factors. Despite substantial progress in both swarm robotics and human factors research, relatively little work has explicitly pursued an integrated perspective that considers how human cognition and the swarm's distributed nature should be effectively combined. To address this shortcoming, this thesis adopts a joint-agent perspective that considers the human and the swarm as parts of the same intelligent, scalable, robust, and flexible agent. More specifically, this thesis (1) presents the first detailed taxonomy focusing on HSI in the context of the swarm's distributed nature; (2) formulates Extended Swarming (ES) as a bio-inspired, integrated design approach to HSI; (3) develops and analyses agent programs for ES; and (4) investigates, through a user study, the sense of agency and the sense of ownership over swarm bodies. Firstly, this thesis specifies the problem of HSI from both the swarm and the human perspectives. An analysis of swarm embodiment within HSI reveals different levels, showing that (1) a human operator should influence the world-embedded self-organisation of the swarm for swarm control, and (2) the swarm should itself provide global perceptual information to the human operator. A taxonomy structures the design space of HSI, which demonstrates its complexity. On this basis, this thesis extracts design concepts from the literature to reduce design space complexity. Secondly, this thesis formulates ES as its main contribution, a bio-inspired, integrated design approach to HSI that explicitly considers both human cognition and the swarm's scalability, robustness, and flexibility. Drawing on similarities between the hierarchical neural control architecture and HSI, ES treats the swarm as a controllable, embodied extension of the human operator. ES integrates the human into the swarm body as its "brain", exerting high-level modulatory control, while the swarm body reacts to environmental situations in a semi-autonomous manner, without direct operator involvement. From the human's perspective, ES presents the swarm as an extended part of the human body. Thirdly, this thesis develops and evaluates agent programs for: (1) swarm body control; (2) global perception; (3) semi-autonomous adaptation; and (4) perception tuning in an abstract search and rescue simulation, in which the human operator searches for victims using their swarm body. The result is an intuitively controllable swarm body that extends the operator's sensorimotor range substantially over 360°, while embedding a neural connectome. By treating the swarm as an embodied vector field, ES enables control through intuitive, scale-invariant parameters that shape the swarm body's macro-level behaviour, such as the body's radial stretching range. The swarm body provides global perception by identifying and stabilising robot chains inside the swarm body that transmit environmental information and terminate at the centre of the swarm body. ES achieves swarm body adaptation by integrating swarm body control and perception through a reaction function that depends on both the global swarm body percept and human state signals, analogous to the hierarchical architecture of the nervous system. Treating robots as embodied neurons enables enactive perception tuning, thereby allowing intuitive control over perceptual properties such as the swarm body's sensitivity to external stimulation. Fourthly, this thesis investigates the human factors requirements for swarms as controllable embodied extensions of humans. A literature overview reveals potential requirements for the sense of agency, referring to the feeling of controlling the swarm body, and the sense of ownership, referring to the feeling that the swarm body belongs to the human body. A user experiment investigates the sense of agency and the sense of ownership as a function of three different control input channels: (1) joystick control; (2) body posture control; and (3) heart-rate-based control. While body posture control proved to be as effective as joystick control in terms of the sense of agency, the experiment revealed no significant effect on the sense of ownership. The study provides a baseline for future investigations into designing robot swarms as controllable, embodied extensions of humans. These contributions together establish ES as a novel, integrated design approach to HSI, and demonstrate its utility in guiding HSI design through the principled development and analysis of specific solutions across both swarm robotics and human factors engineering.
DOI: 10.48565/bonndoc-948
Institution: University of Bonn
Language: English
Field: Psychology
Open Access status: green
OpenAlex ID: W7204171051
Cites:
Cited by:
Related to: