Loud humming sounds, clattering machine-like noises, focused looks and mumbling voices. TheDepartment of Automatic Control’s robotics lab, which is an educational and research facility shared with the Department of Computer Science at LTH, is a meeting place for students and researchers not only from Automatic Control and Computer Science, but also those from many other academic disciplines, such as mathematics, production techniques and cognition research.

“We all collaborate in developing the robots of the future”, says Anders Robertsson, professor at the Department of Automatic Control, and head of the robotics lab.

With the exception of a few robots developed specifically to evaluate new conceptual approaches, most of the robots in the lab are industrially available robots that have been reconfigured for experimental purposes, but that is where the similarities end. One robot has been developed to improve the assembly capacity in industrial production, and another to build houses substantially faster and more efficiently than any human can. The purpose of a third robot is to improve medical care, by filming from the surgeon’s perspective during an operation.

“Operations can be very long and demanding, and it is not practical for surgeons to wear heavy camera equipment on their head for such a long period of time. Our robot helps the surgeon by carrying the equipment. The camera “floats” above the head of the surgeon, following his or her slightest movement, and filming every detail of the operation as seen through the eyes of the surgeon”, says Anders Robertsson. This allows operations to serve as learning platforms, not only for the other members of the team assisting the surgeon, but also for medical students or surgeons preparing for similar types of operations.

Autonomous and collaborative systems underway

The medical robot is part of a current shift in the focus of the robotics lab, from traditional industrial robotics to more autonomous and collaborative systems.

“We’re working on letting robots handle uncertainties, which would allow them to operate in completely new environments and settings, such as fixtureless and unstructured environments”, says Anders Robertsson. In order to do so, researchers utilize sensory feedback, not only in the form of vision but also tactile sensing.

“It’s all about making a change, and measuring the reaction, and then making another change and measuring the reaction, and so on, until we achieve the desired reaction or behavior.”

Another approach is to create a dialogue system for communication between robots and operators, so that the operators do not have to be skilled software programmers to instruct robots efficiently. Assembly has always been considered a promising robotic application, but it has proven difficult to automate it in reality due to complex time-consuming programming and the need for dedicated hardware. As a result, large volumes of expensive products are still assembled manually in low-wage countries under poor conditions. However, a system was developed in a recently completed EU project called SARAFun [länk: cordis.europa.eu/project/rcn/194263/factsheet/en], which allows non-skilled users to teach a collaborative robot a new bi-manual assembly task in less than a day.

“The overall concept is that the robot learns and executes the assembly task in a human-like manner”, says Anders Robertsson.

Learning by observing

Initially, researchers studied human assembly workers in order to understand how they learned and performed assembly tasks. The human performance was then modeled and transferred to the robot as assembly skills.

“The ultimate goal is that the robot will learn the assembly task, such as insertion or folding, solely by observing the task being performed by a human instructor, or by the instructor showing the robot what to do, for example, by drawing a picture. The robot would then translate these visual instructions into a new movement pattern.”

The robot’s understanding of the assembly task gradually improves, thanks to its cutting-edge sensory and cognitive abilities, and the human robot interaction, until it finally learns to perform the task quickly and reliably.

“If we succeed in offering a standardized hardware setup and simple fitting of the collaborative robot into existing workplaces, we can keep production closer to home and thus reduce transportation. This would be beneficial from both the environmental and economic perspectives”, says Anders Robertsson.

The effect of the widespread introduction of industrial robots on employment figures does not worry him.

“I believe new tasks will be created for the workforce alongside the ones performed by machines”, he says, referring to recent statistics published by the International Federation of Robotics.

Robotics lab soon to be an open-door lab

Although collaboration has always been central to the robotics lab, it will soon enter an even more cooperative phase, by becoming the tenth “open-door lab” at Lund University. This means that anyone (mainly companies) can rent the robotics lab to perform their own experiments, using the state-of-the-art equipment, facilities, networks and, last but not least, the competence of the researchers at the lab.

“We are very pleased with this development. Industrial cooperation has always been important within robotics research, and we are always looking for new and exciting collaborations”, says Anders Robertsson.