Ethical Consideration of Machine System Control in AR research

When considering pursuing research into using “Extended Reality Systems” (XR) as they interface with robotic and automated manufacturing components, many ethical concerns need to be considered before designing and implementing research methods. Human health will need to be protected as well as any personal or biometric data collected.

In this analysis, we will assume that the research is in the planning phase and will use human subjects directly controlling robotic equipment on a factory floor through augmented reality goggles. Although there are many concerns about ethical behavior in this study, here we will only discuss the most relevant concerns for the human subjects; a safe testing environment and protocols to secure personal data.

First and foremost, the testing environment for human subjects must include protocols to reduce the risks of injury to all participants. Because this type of research includes human test subjects and the collection of personal data, IRB review and approval are essential to developing and implementing safety protocols. These protocols would be vetted and developed through the IRB process using established human subject research guidelines.  Possible safety protocols include; wearing proper safety equipment, training for the safe use of the research tools, risk management training, machine overrides, physical lockout/cutout equipment and machine guarding. Through proper training and protocol, exposure to safety risks can be mitigated in the research facility.

For the AR command and control metrics, personal data will be needed to customize the user interface of the AR systems. Metrics such as height, weight, inter-ocular distance, arm reach, hand size, and the ethnographic background will be collected. The IRB rules on collecting and managing this data are well established. What isn’t so well defined, however, is the other biometric data that some of these emerging technology systems collect, store and use in the background. To subjugate this possible oversight, a deeper understanding of the “unpublished” data points these emerging technological devices collect data will need to be examined.  For example, does the volumetric video that a headset uses for localization contain any biometric data of other items or humans within the frame? Can sensor data be extrapolated to determine the identity or location of the test subjects?

 In the proposed use case scenario, biometric and location-based data outside the scope of the research may be generated, passed, and stored in the cloud for processing by Microsoft Azure Connect content management and analysis systems. These cloud structures act as a storage system, interpreter, and database generation for the data collection points. This data then interacts with AI algorithms to generate information for the user and researcher. Therefore, the collection strategy for these unintended data points should protect the test subjects throughout the research process from collection to publication.

Other restrictions, such as those of the sponsors or government agencies, should be considered, and these restrictions should be relayed and adhered to by the vendors processing the data. In this case, provisions between Microsoft Corporation and the researchers should ensure that great care is taken to protect the privacy of the test subjects and intended/unintended data generated from the research. Protocols between all interested parties will also need to be established to secure the research data well beyond the project’s initial timeline.

With proper care of the human test subjects, agreements with cloud vendors, and a deeper understanding of how and where users biometric and survey data will be used, protections for research can be established that adhere to modern ethical research and safety standards.