Robots and Communication

E. Sandry

• Publisher: Palgrave Pivot
• Year: 2015
• Print ISBN: 9781137468369
• eBook ISBN: 9781137468376
• Topic: Social Sciences
• Subtopic: Computer Graphics
• Index: Social Sciences

This book explores communication between humans and robots. Using a range of communication theories, it highlights how each theory provides a different perspective on the communication that occurs. The analysis of human interactions with a variety of forms of robot suggests new ways to perceive what communication, and being a communicator, entails.

Part I

From Data to Dogs

1

Designing Robots to Communicate with Humans

Abstract: In Chapter 1, the pursuit of humanlike form is analysed in both fictional and real-life contexts. Amongst other justifications, people committed to building humanoid robots argue that these robots are best suited to work in human environments and to communicate with humans. Two paths in humanoid robot design are considered, but both involve understandings of communication that value commonality over difference. This chapter draws on the critiques of the pursuit of sameness found in the work of communication scholars, which is not much discussed in robotics, to destabilise the assumption that humanlike form is the best form for communicative robots.

Sandry, Eleanor. Robots and Communication. Basingstoke: Palgrave Macmillan, 2015. DOI: 10.1057/9781137468376.0006.

When designing robots to communicate with humans, roboticists draw on a number of different arguments for pursuing a level of human likeness in their designs. Some of these arguments are based on assumptions relating to the practical issues of making a machine that can operate in human-tailored environments. However, for some robots a further concern is to make the robot communicate in familiar humanlike ways. Achieving this design goal is not easy, but the goal itself is strongly reinforced by the widely accepted understanding that communication is based on what communicators have in common, with its success being related to increasing that commonality further. Though not discussed much in the field of robotics, communication theorists have critiqued the pursuit of commonality, and their arguments are explored in this chapter as a way to question the idea that humanlike form is the best path to adopt when creating communicative robots that might also be able to collaborate and work with people.

Science fiction origins

As was discussed in the Introduction, the word robot was used for the first time in Capek’s play R. U. R. to refer to artificially created entities made from organic components and described as almost indistinguishable from humans. When staging the play, the robots are played by human actors who follow Capek’s stage directions by making the robots appear ‘slightly mechanical in their speech and movements, blank of expression, fixed in their gaze’ (1920/2006, p. 5). The portrayal of the robot characters emphasises that they have no emotions or feelings, not even sensing pain in the early stages of the play’s narrative. In spite of their rather staid behaviour, Rossum’s robots still retain their humanlike abilities to act upon and respond to spoken orders and gestures, as well as being able to read written instructions. The original robots are therefore presented as the ultimate workers, easily instructed and able to perform all manual tasks that are asked of them without ever complaining.

The humanlike form of Rossum’s robots not only explains the ease with which they can replace human workers in human-tailored environments within the play, but also supports Capek’s use of them as figures that highlight the social and cultural divisions between factory workers and the privileged ruling class in his own society (Philmus, 2005, p. 103). The name robot, derived from the word ‘robota’ meaning ‘forced labour’ in Czech, positions these emotionless artificial beings as the servants or slaves of humankind and emphasises Capek’s message that ‘the industrial system treated human labourers as though they were machines’ (Disch, 1998, pp. 8–9).

After this early introduction, the robot went on to become a familiar icon in science fiction, although more often made of metal than organic material. Narratives continued to concern robots as workers and perpetuated the assumption that humanlike form would enable robots to carry out what were once human tasks effectively, as well as supporting easy communication with people. This is notably the case in the fictional work and speculative essays of Isaac Asimov, a scientist and writer whose conception of robots and robotics remains influential today. Although Asimov did write a number of stories about non-humanoid robots, an example being the robotic cars in the short story ‘Sally’ (1953/1995, pp. 19–40), he was fascinated by the idea of making machines in human form. In his nonfiction essay, ‘The Friends We Make’, Asimov explains that his narratives are concerned largely with the human desire for a servant of a similar level of intelligence, a worker who is tireless, strong, contented and never bored (1977/1990).

In spite of the differences in their construction, there are clear similarities between Asimov’s mechanical robot characters and the fictional Rossum’s artificial humans. In Asimov’s opinion, as in Rossum’s, humanlike form and an average-sized body is an essential attribute to enable a robot to operate as an effective worker in an environment already tailored to humans (Asimov, 1977/1990, pp. 417–419; Capek, 1920/2006, p. 12). It is worth remembering that Capek’s play dates from 1920, and Asimov originally wrote his essay in 1977, which may explain why the idea of making human environments and workplaces more accessible to all human individuals including, for example, those people using wheelchairs, was not considered significant by either author. In contrast, present-day requirements to address accessibility concerns for humans also allow wheeled or tracked robots to access those spaces more easily.

Humanoid robots in real life

In spite of the increased accessibility of present-day workplaces, Asimov’s conception of robots as humanoid machines constructed from inorganic materials, and the idea that their humanlike form lends itself to working in human environments, is mirrored in the creation of real-life robots and discussions about their designs. For example, the design and size of Honda’s bipedal walking machine, ASIMO, is understood ‘to allow it to operate freely in the human living space’, so that it can ‘operate light switches and door knobs, and work at tables and work benches’, while also making it ‘people-friendly’ (Honda, 2007, p. 15). The development of ASIMO has taken over a decade, and most of that time has been spent trying to perfect its walking, running and stair-climbing abilities. Publicity information regarding ASIMO suggests that its design enables it to work freely in unmodified human workplaces and homes. However, its ability to operate in constantly changing real-life environments is still fragile, and while Honda’s websites contain videos of ASIMO’s successes, a number of amateur videos on the Internet record some of the moments when ASIMO has experienced difficulties, including, for example, falling on stairs.

Towards the end of his essay, Asimov says that if people are to accept ‘thinking partners – or, at the least, thinking servants – in the form of machines’ it is likely that they ‘will be more comfortable with them’ and ‘will relate to them more easily, if they are shaped like humans’ (1977/1990, p. 419). He extends this concept even further, as he considers the possibility of humans living more closely with robots, suggesting that ‘[i]t will be easier to be friends with human-shaped robots than with specialized machines of unrecognizable shape’ (Asimov, 1977/1990, p. 419). This appraisal is supported by the way that ASIMO’s designers express their desire to make a robot that is people-friendly. It is clear that both Asimov and present-day designers of robots base the drive to make humanoid robots on factors that relate more closely to concerns of sociability, and the need to make robots more personable, than simply enabling them to work in human physical environments. In the course of this development, a rather uncomfortable juxtaposition between the idea of ‘robots as servants’ and ‘robots as friends’ arises, exacerbated by the origin of the term robot with Capek’s concerns regarding factory workers and the idea of forced labour. These incompatible perspectives are still present in many discussions of present-day robot designs.

The idea that ASIMO needs somehow to seem friendly results in the way that this robot is designed to be slightly smaller than the average adult human and thus nonthreatening and also underlies the expressiveness of its communication. While ASIMO’s designers have in the main concentrated their efforts on solving the problems of bipedal movement, they have also made ASIMO an adept mimic of many human hand and arm movements. ASIMO waves at people and shakes their hands, demonstrating how its humanlike form has been tailored towards supporting friendly communication with humans, as well as enabling its movement around the world. Although ASIMO’s hand and arm movements show a tremendous potential for emotional expressiveness they are not overtly used to express emotional content. Indeed, there is no discussion of emotion as important in ASIMO’s character on the Honda website, and in all versions its ‘face’ has remained immovable and partially obscured by a tinted visor. Therefore, although ASIMO may say things that contain some emotional content, its overall design demonstrates the idea that the expression of emotions is not that important to the operation of the robot or its interactions with humans.

Emotional robots

This raises the question of whether robots can, or should, express humanlike emotions in their communication with people. Even in Capek’s play, the relationship between the conception of the robot, and the potential for its ability to feel, or to express emotions, was complicated. As the narrative of R. U. R. progresses, the robots are given the ability to sense pain, framed as a practical design decision in order to help them avoid damaging their fragile organic bodies as they work. However, towards the end of the play a range of emotions begin to emerge in the robot characters and, although the details are left rather unclear, it does transpire that a few robots were built with a ‘level of irritability’ as a way of ‘making them into people’ (Capek, 1920/2006, pp. 61–62). Unfortunately, while the robots’ human designers may have meant to create more friendly robots, in the conclusion to R. U. R. the robots rebel against their human masters and it is their ability to relate to one another emotionally that allows the robots to survive and replace humanity in the world.

Asimov’s fictional descriptions sometimes imbue robots with underlying feelings, although they often have difficulty expressing them as emotions to others. In ‘The Bicentennial Man’, for example, Asimov describes the robot Andrew’s feelings as hidden because of the ‘smooth blankness’ of his face for much of the narrative (1976/1990, p. 245). It is only at the end of this short story that Asimov integrates Andrew’s inner feelings with his expressions when upgrades to the robot’s face make its features mobile, allowing the robot to smile (1976/1990, p. 290). It would seem that though Asimov’s speculative essay ‘The Friends We Make’ does not overtly value the ability of robots to express emotions in their communication with people, in spite of positioning them as potential friends, his stories constantly question this idea. ‘The Bicentennial Man’, in particular, positions emotional expression in terms of body movement, tone of voice and facial expression as an important part of human-robot relations and communication.

The narratives of R. U. R. and ‘The Bicentennial Man’ highlight a difficult relationship between the robot as a constructed being, its ability to express emotions and its capacity to feel. The way that robots move, communicate and complete tasks places them as ‘thinking’ and also potentially as somewhat ‘alive’ but, as Turkle has noted is the case for computers (2005, p. 63), early in both stories humans are identified as unique because they have ‘an emotional life’ that robots lack. By the end of Capek’s play and Asimov’s short story this distinction is dissolved, the emotional development of Rossum’s robots being driven by a human desire to make the robots more like people, whereas many of the changes to Andrew are chosen by the robot himself. In both cases, the robots are described as more richly communicative as a consequence.

It may therefore be unsurprising that some roboticists set aside the problems of perfecting bipedal movement to focus instead on maximising the expressiveness of the robot’s face. Broadly, there are two different paths that these designs follow. The first leads to the creation of humanlike robotic faces that are virtually indistinguishable from humans, bringing Capek’s artificial humans to mind. The second involves the creation of robots that operate more like caricatures of humans, a development path discussed later in this chapter. In terms of realistic-looking humanlike heads, the Japanese roboticist Hiroshi Ishiguro is famous for creating his own robot double, whereas David Hanson has created robotic heads that are accurate representations of Phillip K. Dick and Albert Einstein, as well as generic humanoids like Jules, built for the Bristol Robotics Laboratory (BRL) in the United Kingdom. Hanson explains that his aim is ‘to model the behavior and movements of people in robots that act and react virtually indistinguishably from their human counterparts’ (Hanson Robotics Website). For both Ishiguro and Hanson, the design of robots that communicate using human language and humanlike facial expressions is key.

Whether they have been designed to move in humanlike ways, or with humanlike expressive faces, very few real-life humanoid robots are sufficiently flexible and robust in their operation to enable their introduction into everyday situations. A few robots have entered the workplace, versions of Saya (a seated robot with a humanlike face) having worked as a receptionist, school teacher and greeting guests in a Tokyo retail store (Hornyak, 2009). ASIMO has also recently been trialled as a museum guide, although not entirely successfully (Hornyak, 2013). However, most humanoid robots are still found in laboratories and other spaces where people’s interactions with them are carefully framed and controlled. It is therefore only in fiction that this type of humanlike robot has reached a level of sophistication that enables a more complete interrogation of the perceived benefits of such designs. In particular, Lieutenant Commander Data, the android officer in the television series Star Trek: The Next Generation (ST:TNG) and related films, provides a paradigmatic example of one possible future of this pathway in humanoid robotics.

Communicating with Data

Data is portrayed as a highly sophisticated humanoid machine, able to gesture and move in more humanlike ways than ASIMO, with a face that is capable of mimicking human expressions even more precisely than that of Jules. As his name suggests, Data is best regarded as a technologically advanced, embodied computer system. This robot’s ‘brain’ is shown as composed of silicon chips and other electronic components, and he runs self-diagnostic programs when he feels there may be an error in his circuitry.

In terms of his appearance, Data is very humanlike indeed, although the makers of ST:TNG have defined some physical anomalies, such as his rather unusual eye colour and the metallic sheen to his skin, the latter being particularly noticeable in the first series. In addition, Data has some quirks in his behaviour that set him apart; for example, he rarely uses verbal contractions, saying ‘is not’ instead of ‘isn’t’. The difference between Data and humans most explored in many of the storylines is his lack of emotion, or more correctly the underlying feelings associated with emotional responses. Again, the discussion of Data’s feelings resonates with Turkle’s appraisal of the ‘emotional life’ that computers are understood to lack (2005, p. 63). However, Data, as an embodied computer with a humanlike face and body, is able to express the appropriate human emotional responses to different situations, although his inability to experience the related feelings continues to set him apart from the other crew members of the Enterprise.

For many of ST:TNG’s storylines, it is important that Data is almost indistinguishable from humans, while also remaining somewhat other-than-human in appearance, behaviour and ability. This subtle difference results in occasions when human characters are shown forgetting that Data is different from them in various ways. For example, in one episode Commander Riker makes several unsuccessful attempts to open a jammed door, before eventually stepping aside to allow Data, with his superior strength, to open the same door with ease (‘Hide and Q’, ST:TNG, Season 1, Episode 9). In addition, Data’s human appearance and expressive face means that he often has to remind other characters that he does not have any feelings to hurt. In spite of Data’s differences from humans, he experiences very few difficulties communicating with, working amongst and being accepted by the human, and humanoid alien, members of the ship’s crew. Eventually, with the help of the ‘emotion chip’, Data is able to feel the effects of his emotions, a development that is portrayed as fraught with difficulty in its early stages as he learns to control the feelings that threaten to overwhelm his ability to function.

Most of Data’s responsibilities in Star Fleet are concerned with research, information retrieval, processing and analysis, as might be expected in his role as Science Officer. An important part of such work, as is true for human scientists, is the ability to communicate results and conclusions to others. In general, Data seems to have few problems in this respect, although he is often unusually precise in what he says, using a systematic approach in his interactions and showing a particular concern to provide detailed and accurate information. The portrayal of his communicative acts stresses the importance of factual information, logic and objective rational argument. Data’s communication style can therefore be analysed from the perspective of cybernetic and semiotic traditions of theory, which focus upon the accurate transmission or exchange of information, supported by precision coding and decoding in language.

The cybernetic tradition has its foundations in the work of scholars such as Claude Shannon and Warren Weaver (1948), Norbert Wiener (1948) and Alan Turing (1950), being drawn out of and feeding back into research into systems and information science, artificial intelligence and cognitive theory. The driving philosophical assumptions behind this tradition are concerned with materialism, rationalism and functionalism, the result being theories that regard all communication in terms of information processing and exchange within systems (Craig, 1999, p. 141). Of course, Data’s construction as an embodied computer system serves to link his design with cybernetic theory, and with what is known as the cybernetic tradition of communication, even more closely.

As Craig notes, the cybernetic tradition shares some common ground with the other communication traditions he identifies, including the semiotic tradition. For Craig, this similarity is based on the way that semiotics collapses ‘human agency into underlying or overarch...

Table of contents

• Cover
• Title
• Introduction
• Part I  From Data to Dogs
• Part II  Communicating with Non-Humanoid Robots
• Part III  Rethinking Robots and Communication
• Conclusion
• Bibliography
• Index