| Event Information
Lockheed Martin Robotics Seminar: Variables, Measures, and Metrics for Autonomy
Variables, Measures, and Metrics for Autonomy: From driverless vehicles to robotic dogs and everything in between
Kristin E Schaefer-Lay, PhD
Chief, Autonomous Systems Branch
Kristin E Schaefer-Lay, PhD is the Chief for the Autonomous Systems Branch of DEVCOM Army Research Laboratory. Dr. Schaefer-Lay has led research efforts for ARL in the areas of trust, robotics, and autonomy for the Robotics Collaborative Technology Alliance, Applied Robotics for Installations and Base Operations, Army Wingman Program, and Human-Autonomy Teaming Essential Research Program. She now leads a team that specializes in robotics and systems engineering across wheels, wings, and appendages. She has over 70 publications and 90 presentations in this and associated areas of research. She holds a PhD and MS in Modeling & Simulation from the University of Central Florida, and BA in Psychology from Susquehanna University.
Developing effective and appropriate autonomy for robotics, whether ground, air or limbed, is a hard problem. As robotic systems move beyond pre-programmed automated behaviors to adaptive real-time autonomy (tools to team members), the complexity of this problem multiplies. The number of variables that can, and often do, influence the functionality of each algorithm are often not even considered during the initial development of a specific behavior. Further, how to measure “success” or “failure” is often mistakenly thought of as a final or close to final step in the development cycle. So how do “you” account for the variables that you don’t know? How do you measure something that doesn’t exist…yet? This interactive presentation will provide lessons learned approach working from the laboratory, to simulation, to field and back again with examples from our work on driverless vehicles, ground vehicles, limbed robotics, and small unmanned aerial systems. You will walk away knowing some…but clearly not all…of the common mistakes researchers make when developing autonomy, tricks of the trade on simulation systems, and an effective development lifecycle for robotics development, including processes for identifying appropriate variables, metrics, and measures.