Consumer Robotics

Roboskel develops the technologies that will enable our vision for consumer robotics where complex robot tasks can be defined, as opposed to selected, by end-consumers with no technical expertise: we envisage a human-robot interaction technology that is far more versatile and dynamic than selecting among a predefined menu but at the same time is restricted and unambiguous (that natural language is not) and more natural and intuitive than developing software. However, where "natural" and "intuitive" in HRI are conventionally understood to equate to spoken interaction, we argue that people have been exposed and accustomed to a variety of different interfaces with machines: from cars' steering wheels, shift sticks and pedals, to gamepads and TV remotes there is a huge selection of ways to unambiguously and yet naturally and intuitively "explain" to a machine what needs to be done without uttering a single word.At the core of our approach is perceiving robot actions as a matching of concrete items from an inventory of robot capabilities against specific affordances offered at different locations in the robot’s environment, and perceiving controling robots as sequences of such actions conditioned by perception. Our plan is to develop an environment for controlling robots that follows the visual programming paradigm for authoring visual representations of a system of actions and reactions to external stimuli; and that allows for new capabilities to be demonstrated and added to the library and new affordances to be pointed out to the robot via a combination of teleoperation and HRI.This augments visual programming with opportunities offered by the robot's embodied nature: the ability to adapt by receiving external reinforcement stimuli and the ability to be instructed by being teleoperated. We believe that embodied visual programming will be a decisive step towards consumer robotics.

Results up to now include experimenting with different teleoperation controls implemented for ubiquitous mobile devices such as smartphones and tables (HRI 2014) and setting an applied research plan for deploying consumer robotics in health monitoring applications (PETRA 2013).