Abstract
The Ukraine war shows how quickly commercial FPV drones and unmanned maritime systems like MAGURA can shift from civilian or experimental technology into weapons. In this sociotechnical research paper, I explore the accountability dispersion in increasingly autonomous systems adapted for warfare. I also address the dual-use problem, which refers to technology intended for civilian use being adapted for combat, and how this impacts proliferation. The paper includes a thorough analysis of the social construction of the MAGURA drone; a Ukrainian naval drone vessel deployed in the Russia-Ukraine war of 2022. In addition, the legal implications and historical interpretations surrounding autonomous systems, specifically drones, for warfare are also cross analyzed to acquire a fair and complete understanding. As an analysis framework, the Social Construction of Technology (SCOT) is used to examine how these autonomous systems take shape, and Actor Network Theory (ANT) is relied on to examine how accountability is distributed between human and non-human actors once those systems are ready to be deployed. Through this analysis, I dissect the social construction of the MAGURA series, proving that these developments are not solely driven by technological progress.
In the technical report, my team and I, describe our capstone project involving a deception-resilient robotics security system that attempts to integrate Wi-Fi-based sensing, and multi-robot motion planning for intruder interception in indoor environments. The system uses two robots that are programmed to cooperate, predict, and constrain a third robot programmed as a deceptive intruder. A Wi-Fi channel state information sensing pipeline is configured for intruder localization. However, due to the novel and unstable nature of live detection, a motion capture system is used to emulate outputs from the Wi-Fi sensing module for testing purposes. In any given scenario, one robot chases and influences the escape route of the intruder by blocking exits and using smart positioning techniques. Meanwhile the second robot strategically positions itself to corner and trap the intruder based on the updated actions of its partner robot. The system supports variable indoor layouts, due to an adapted A-star algorithm. Through different trial runs, the system is able to successfully capture the intruder with satisfactory rates of above 80%.
The STS thesis and technical projects, both cover autonomous systems in different contexts. As previously described, the robot security system is designed for protective indoor usage. However, it still relies on traditional sensors, prediction, autonomous motion planning and is built on a highly adaptable Turtlebot4 platform. This leaves it prone to the dual-use problem that is described in the STS thesis paper. The Ukraine case study within the STS thesis paper demonstrates that systems framed as civilian-bound, and experimental can be easily reproposed for combat, and other non-intended usages depending on social groups and environmental pressure. Thus, I argue that these technical systems must be treated and understood as actors within a bigger sociotechnical network with different actors that can interact. In the context of MAGURA, this sociotechnical network involves designers, users, government institutions, and different components of the drone itself as the non-human actors. The technical project and STS thesis project reveal that human actors involved in the development of these systems must be aware of these realities and design with accountability, transparency, and misuse prevention in mind.
