Robot from TU Wien learns to clean sinks

From around 3,500 papers submitted to the ‘IROS’ robotics conference recently held in Abu Dhabi, the work of researchers from the Institute for Automation and Control Engineering (ACIN) of the Vienna University of Technology was selected for a ‘Best Application Paper Award’. and was therefore selected by colleagues as one of the most important innovations of the year. This shows that “useful things” are being done in Vienna in this “scientifically very dynamic area worldwide”, explain Andreas Kugi and Christian Hartl-Nesic of ACIN.

Learning with a sponge
In this specific case, the AI ​​system learns more or less directly from humans. The vehicle for this is an ‘instrumented tool’ – here a standard household sponge, which the scientists quickly punctured and fitted with force-torque sensors. In this way it is recorded in detail how much force is used when handling the sponge, at what point and at what curvature of the workpiece – in this case the porcelain basin.

Watching is enough
In addition, the movements of the leading hand are accurately recorded in 3D via an optical tracking system. In combination, you get hundreds to thousands of measurement data per second. This very high resolution also explains why the downstream data processing process, which is based on artificial intelligence or machine learning, can get a reasonably good idea of ​​the requirements after just a few seconds of ‘looking’. “We can understand the process very well,” as Hartl-Nesic put it.

Behind the ‘learning’ is a mathematical-statistical approach that can be used to break down similarities in movements. This means that what has been shown can be technically translated and ultimately simulated in the form of very few parameters, the scientist explains: “For each new edge, the system can use the statistical description to calculate a meaningful tool movement.”

The demonstrations showed that the sink task can be replicated after minimal demonstration – a person simply wipes the front of the sink for a few seconds – and can also be applied to new situations, such as different curves. In other words: the rest of the geometrically very demanding object is then cleaned automatically. In the future, the robot should also check its success after performing the process to rework it if necessary.

Keep waiting
However, this assistant will not quickly find its way into households: the high-tech robot arm and the advanced system from Vienna are not available ready-made in this form. Kugi: “A lot of fundamental research still needs to be done here,” but the ingredients are there.

The researchers’ vision is in the direction of ‘robots as assistants’. Kugi was convinced that this is especially important in the medium term for Austria as a country with many small and medium-sized manufacturing companies in the craft sector – also in view of the increasing shortage of skilled workers. Such systems would “not replace people, but support them en masse”.

Area of ​​application: Almost everywhere where various types of surfaces need to be sanded, polished, painted, glues need to be applied or parts need to be assembled.

Now is the time for companies to think about what their production will look like in ten to fifteen years. ‘Intelligent assistance systems with semi-autonomous functions’ are part of this. It would also be conceivable to gradually and collaboratively improve such approaches across multiple companies and applications: “Private data – for example about the specific shape of a particular piece of work – would remain private, but important basic principles learned would be exchanged to further improve the data improve the capabilities of all robots,” says TU Vienna.