BY NATHAN KIRCHNER
If someone tells you to think of a robot, what springs to mind? Is it a humanoid shape made of metal, with glowing eyes, that speaks in a jerky voice?
Or is it a robotic factory arm, or a car that can park for you, or maybe a system that heats or cools your house?
For some time now pop culture has painted a particular picture of robots. From Asimov’s ‘bots, to the Terminator – even the Transformers – the very concept of a robot has grown up next to these hugely popular sci-fi characters.
So why aren’t we seeing robots like these by now?
On one hand we have the glamourised depictions of sci-fi robots. On the other we have the roboticist’s more pragmatic view of robots as machines that perform functions in an autonomous way.
For these roboticists (myself included), robots are all around us – in our cars, our homes, on public transport and in buildings.
Part of the problem is that there are a number of research projects around the globe that seem to fit the “Hollywood” robot image – Hiroshi Ishiguro’s Intelligent Robotics Laboratories with their Geminoid and Robovie enabled research and our own RobotAssist to name just a couple.
This is one major source of the confusion surrounding the state of the art in robotics. In an attempt to make our research accessible to the wider world we, the roboticists, have leaned on pop culture’s sci-fi robot and subsequently reinforced the stereotype.
Sure, the fundamental research questions we are probing are embedded in the project, so there’s no harm done, right? Well, yes actually, there is.
We’ve shaped the presentation of our research around this stereotype and the actual science questions are less visible to the casual spectator of our work.
With robotics research and development presented in this manner, the tendency for the casual onlooker is to measure the gaps between the research on show and the benchmark of the sci-fi robot. This is not always a true indication of the state of the art.
I myself have been guilty of unintentionally obscuring my own research intentions by putting them in a sci-fi friendly wrapper with RobotAssist. While RobotAssist appears on the surface to be another somewhat human-like robot that can do some cool things (but is no T1000), this isn’t its intended role.
RobotAssist is a research and development platform for core robotics technologies. It has provided a valuable platform for a number of important developments that have found their way into real-world realisations. These include robust people-detection and tracking techniques that are currently deployed in mining, construction and transport environments.
From a particular robotics perspective there is little difference between the RobotAssist incarnation of the technology and how it is used in a transport environment. What changes is the way the technology is embodied.
Your local train station, say, doesn’t look anything like a robot. But in a sense, the entire building is a robot. Maybe the security cameras and embedded sensors are its eyes, maybe turning on and off exit signs and dynamically restricting and redirecting some passageways are the actuators. This kind of robot is invisible.
Put your body into it
Roboticists have been guided by the “sense-act-think” operational definition of a robot for more than 30 years now.
This definition states that a robot is a machine that can actively “sense” the state of the world, “think” intelligently about its task in light of sensed information to form an action plan, and “act” that plan upon the world.
Notice there is no mention of embodiment?
This, I believe, drives another major source of the confusion surrounding the state of the art in robotics.
Society is conditioned by pop culture to recognise robots through the way they’re embodied. Roboticists, however, often consider the embodiment superfluous, or at least tangential, to the robot.
This brings us back to the two viewpoints that I mentioned earlier. The reality is that disembodied robots are already prevalent throughout society.
Just think about our cars with their automatic parking (see video above) and braking. We don’t tend to acknowledge these machines as robots, partly due to the sci-fi stereotype and partly due to roboticists further encouraging this stereotype – but they do fit the operational definition of a robot.
Don’t get me wrong, the sci-fi style robot entering society is inevitable. Too many people want it for it to not happen.
My point is that this is just one of the many forms a robot can take, and perhaps it will be one of the later ones to be realised.
If we want a true gauge of where we are at with robotics we may need to re-calibrate our expectations of what a robot is. We are “getting there” with our research. It just turns out that “there” isn’t exactly where pop-culture told us where we should be.