Modeling Human Performance for Human-Robot Systems

Abstract

Human performance plays an important role in system performance and is influenced by a wide set of internal, organizational, and environmental factors. Human-robot systems are being developed for a large number of domains, including in-home care, military deployments, emergency response, industrial robots, and therapeutic applications. Human performance modeling provides the opportunity to investigate multiple system prototypes and theories regarding human performance that can be further evaluated in human-in- the-loop experiments, to quickly adapt to rapidly changing robotic technology, and to represent human behavior in extreme conditions. Interacting with robots also influences human performance, and models of the human-robot system provide an avenue to investigate the impact on human performance that the human-robot teaming elicits. This chapter provides an overview of examples representing common methods for assessing human performance in human-robot systems. The topics are summarized in Table 3.2 and include a description of human performance modeling, a summary of commonly used human performance modeling tools, examples of human performance modeling with a focus on its application in human-robot systems, information regarding the validation of models, and guidelines for implementing human performance modeling techniques for robotics. IntR oduCtIon The performance of humans interacting with a system influences the effectiveness of the system. It is common knowledge that human performance varies and that a number of factors influence human performance. The vast majority of human-robot systems litera- ture over the last decade has focused on evaluating systems and measuring human perfor- mance. It is important not only to measure but also to characterize and model human performance and how it impacts overall system goals and objectives. A critical aspect of future human-robot system design is understanding how human performance can influ- ence interaction with a robot, such that the overall system is more robust in highly uncer- tain and dynamic domains. Modeling of human performance can provide important insight into how to design systems to support human-robot interaction, how to design robots and better interfaces, and how the system can adapt to human performance degra- dations. Human performance modeling has a long history in general system design but has only recently begun to receive attention in relation to human-robot systems.