Collaborative Research: NRI: FND: Grounded Reasoning about Robot Capabilities for Law and Policy

Oregon State University, Corvallis OR

Investigators

Abstract

ABSTRACT This award supports a research project that focuses on developing tools and best practices for promoting effective collaboration between legal and policy experts and roboticists. This collaboration serves two purposes: (1) to better ensure the safe and effective development of collaborative robots; and (2) to create effective and appropriate laws and policies governing their deployment in public spaces. The research team plans to develop a specification language and interactive tools for enabling reasoning about how robots and autonomous systems work, at a non-technical level. Combined with “What If?” case studies, this will facilitate dialog between technical experts and their legal and policy counterparts. The specification language and reasoning toolkit will be designed to evolve simultaneously with, and support, rapid changes in specific technology within the context of existing legal and policy frameworks. The team will use the legal concept of case studies to ground reasoning within a specific scenario. Each case study will consist of a complete description of how the necessary robot capabilities might be implemented, complete with assumptions and potential failures. The results of this research will enable legal and policy experts and roboticists to more accurately reason about the intersection of robotic systems and legal and policy frameworks before either are implemented. Specifically, this will support legal and policy experts in more accurately designing effective policy, legislation, and regulation around robotics that is meaningful and feasible, and will support robotic researchers in designing systems that incorporate legal and ethical questions into their design choices. The research team will engage in an interdisciplinary development of a structured language paradigm (comprised of syntax, semantics, and system descriptions) that supports grounded reasoning about robotics in law and policy contexts. These language paradigms will be instantiated as interactive experiences that enable embodied reasoning and metaphor exploration by experts and non-experts alike. These experiences will be developed around a specific set of near-term, feasible robot tasks, existing robot capabilities and known failure modes, metaphors to explain those capabilities, and specific scenarios that highlight common misconceptions. These experiences will be evaluated on their effectiveness in enabling non-technical people to reason about, and predict, autonomous system behavior. On the legal and policy side, the research team will engage with experts in these areas to develop specific hypothetical case studies (based on the identified tasks) that illustrate how to view technological choices through a law and policy lens. Specifically, team members will work collaboratively to develop a set of associated design patterns that assist in mapping technical concepts to the structured language and the interactive visualizations, develop a prototype suite of interactive training applications, and construct a set of case studies for specific robot-use scenarios that enables meaningful dialog around law and policy and robotics. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.