The field of Human-Robot Interaction is extremely diverse: researchers from Engineering, Computer Science, and Psychology must all work together to produce strong research.
This diversity produces difficulties, as researchers struggle to find common ground for methods and language, often prompting the question of whether the field of HRI is a “field” at all, and whether there is enough commonality to hold together the research community.
We argue here that HRI is a field, but without a universal set of common characteristics. Instead HRI should be viewed in terms of three major areas which contribute strongly to the research and methods used.
RESEARCH AREAS
HRI can be viewed as the intersection of Engineering, Computer Science, and Psychology1. What makes HRI distinct is that all HRI research involves at least two of these general fields. The development of complete systems requires integration of work from each of the fields, but most work occurs within specific intersections.
a) Embodied Cognition (Computer Science & Psychology): One major area of HRI focuses on building computational systems which mimic the cognitive and affective facilities of people. Examples include the intelligence in affective robots which are capable of showing emotion and social robots which are capable of recognizing the emotions and mental states of the people they interact with. Research in this area uses results from Psychology regarding how people understand one another, and also on computational methods for modeling these cognitive facilities. It uses many tools from artificial intelligence and machine learning, but with a stronger emphasis on the embodied nature of the systems (which provides stronger constraints on the inputs of the system, as well as appropriate failure modes and other issues).
An example of work in embodied cognition is that of El Kaliouby [3] and also Rani [8] in estimating cognitive and affective mental states. Another example is that of Gold and Scassellati [6] in developing a robot which recognizes its own body motion.
b) Human-Robot Factors (Psychology & Engineering): The Human-Robot Factors area of HRI focuses on designing robotic systems and understanding how people respond to them. This includes the physical design of robots (such as humanoid vs. non-humanoid designs) as well as the design of control software for robots (as in interfaces for search and rescue robots or military robots). This area may make heavy use of methods from Communications and Human-Computer Interaction studies. One key application in Human-Robot Factors is Urban Search & Rescue [2]. Much of USAR research aims to develop better interfaces for controlling robots. Also in Human-Robot Factors are human response studies such as those of Broadbent et. al. [1]. Human-response studies look to discover what types of robot behavior and design elicit the most desirable social responses from people.
c) Robotics (Engineering & Computer Science): The final area is traditional Robotics, focusing on developing new tools for robot platforms, including both hardware and software capabilities. There is clearly less emphasis on the human (though many robot designs are inspired by people), as the focus is on building the robot and the control software which researchers in other areas of HRI can use. Some common types of Robotics research involve the design of controllers for complex motion (such as walking for legged robots or control of robotic appendages) and development of localization and mapping algorithms.
d) Integrated Systems: There is a fourth area of intersection, where all three areas come together. The goal for HRI is ultimately to develop complete cognitive robotic systems, informed by the principles of Human-Robot Factors, from the ground up. To build these systems requires co-operation between all areas of HRI.
