Martina Philippi: Ethical Aspects of UAV-Based Rescue Robotics

Highlights summarized by Sabrina Breyer

Ethical Aspects of UAV-Based Rescue Robotics

In this Virtual Ethical Innovation Lecture, Martina Philippi, researcher at Ruhr University and researcher at the group “Ethik der digitalen Methoden und Techniken” presents the related project “UAV-Rescue” as a case study for accompanying ethical research in rescue robotics.

Project “UAV-Rescue”

Martina Philippi starts with differentiating the terms “unmanned aerial vehicle” (UAV) also known as “drone” but from a more military context and “unmanned aerial system” (UAS) which comprises the UAV and the ground station. In project “UAV-Rescue” the research is focused on the support of reconnaissance during the so called “golden hour”, when people are likeliest to still be alive in the aftermath of an earthquake. The project goal is to showcase how to increase efficiency of rescue operations by developing a demonstrator that provides a 3D map of the disaster site while localizing persons. The UAV’s main sensors are a radar for vital sign detection and light detection and ranging (LiDAR) for semi-autonomous navigation.

Survey of Ethical Aspects of UAV-Based Rescue Robotics

Focusing on the applied ethics perspective of the project, which Martina Philippi considers to be the “ethically responsible design of the technology (…) mean both the technical design and the design of the later application and research in rescue context” she distinguishes two types of ethical aspects:

“1. Those that are immediately relevant for the development and can be identified together with stakeholders — e.g. safety, data security, privacy” using the methodological approach of the Value-Sensitive Design (VSD) and

“2. Those that don’t appear at first sight have a larger theoretical background — e.g. autonomy of human agency” applying Constructive Technology Assessment (CTA).

In the scope of the first type of ethical aspects, she highlights some results from the first ethics workshops. Here, identified values include “safety, data privacy, security against hacking regarding navigation and data, and the acceptance by responders and victims of the disaster”, which have been gathered with the approach “ethics by design”, the inclusion of end user requirements and the consideration of stakeholder perspective referring to end users, victims of disaster and others.

In the scope of the second type of ethical aspects, Martina Philippi shares some insights from the project. One challenge is the conflict of interest or values regarding its diverging interpretation of (idealistic) ethicists’ and practitioners’ perspectives, which tends to have a preliminary weighting of values due to their commitment to the practical scenario. Here, CTA can be used as a means to alleviate the “Collingridge Dilemma”. Furthermore, she states the challenge of critical use of AI-based technology versus usability aspects which is especially serious in making decisions under stress or triage. Some approaches to solve this conflict in “UAV Rescue” take into account using AI-based technology as a supplement which means to clarify that it is not infallible, it is probably not as exact as it may appear, and it is not taking actual measurements, and thereby counterweighting excessively high expectations.

Requirements for a Responsible Implementation of the UAS

The assurance of responsible implementations in UAS necessitates two conditions: (a) “Applying the “ethics by design” paradigm to the Graphical User Interface” respecting attention guidance, information prioritisation or the best practice idea by having pop-ups for new information, such that it needs not to be shown at all time – having in mind disturbances and potentially unreflected prioritization. (b) “Appropriate implementation of the UAS into the real operation in the heavy USAR (urban search and rescue) context, including training of the responders” which should ensure realistic status and a clear communication of the epistemic status of AI-based information.

Outlook Into a Successful Involvement of the Accompanying Ethical Research

In her conclusion, Martina Philippi highlights some further general requirements for a successful involvement of the accompanying ethical research and thus a responsible development of a novel, AI-supported technology comprising the following aspects:

  • Continuous exchange between stakeholders
  • Knowledge transfer
  • And Openness

In order to meet these requirements all stakeholder groups must be aware of the ethical problems, which is a special challenge in interdisciplinary collaborations in terms of organization, communication and the appropriate mindset.