Collaborative robots have long been able to safely interact with humans thanks to sensors and vision systems. Large industrial robots typically do not offer this capability, which means they can generally perform work separately from humans. New algorithms developed by researchers at the Fraunhofer Institute for Machine Tools and Forming Technology (IWU) change this and enable more secure interaction with industrial robots.
In many a factory, humans collaborate with robots on a product. In most cases, however, this collaboration is relatively limited. For instance, industrial robots are often placed in safety cages to prevent accidents. In other cases, light curtains are used to detect the presence of humans, so that the robots' operation is immediately interrupted if a worker comes too close. These limitations hinder true collaboration, according to Fraunhofer IWU.
Interpreting hand gestures
The institute has therefore developed a number of algorithms that enable more safe interaction with industrial robots. Using these algorithms, industrial robots can recognise human hand and arm gestures, allowing workers to give robots commands. For example, the algorithm suggests a robot tackle a component if it is given by the worker.
Fraunhofer IWU's algorithms look not only at hand and arm gestures, but also at other signals such as body posture and facial expressions. This, according to the researchers, enables robots to interpret gestures better. For example, does a worker look back at a colleague? Then the robot knows that gestures made at that moment are not meant for him.
Recognising dangerous situations
Robots can also recognise whether their movements could pose a danger to human workers thanks to algorithms. For example, does an employee ask the robot to reach for a certain component, but his or her hand is too close to his or her face to do so safely? Then the robot will not carry out the task until the employee's hand is a safe distance away from his or her face.
"We have added effective, safe and flexible interaction to legacy technology. Humans can now communicate and interact directly with heavy robots based on hand gestures for the first time in this industry," explains Dr.-Ing. Mohamad Bdiwi, head of Robotics at Fraunhofer IWU. "Our technology brings motion control to industrial applications. Currently, it is mainly used for gaming applications, such as game consoles."
Quipt
Incidentally, the software Quipt, developed by engineer and designer Madeline Gannon, has been making it possible to interact with industrial robots for some time. Quipt enables robots to recognise wearable markings and respond to them. By sticking these markings on their hand or hanging them around their neck, for example, users can have a robot mimic their movements, among other things.
Gannon demonstrated how it interacted with an ABB IRB 6700 robot using Quipt several years loaded.
Artist gets tethered to robot
A striking demonstration of interaction with an industrial robot is an experiment by performance artist Stelarc, head of the Alternate Anatomies Lab at Australia's Curtin University. Stelarc had himself strapped to an industrial robot and spun around for 30 minutes in 2015. With the experiment, Stelarc wanted to show how intimate human-robot interaction can be.
Author: Wouter Hoeffnagel
Source: Fraunhofer Institute for Machine Tools and Forming Technology (pictured)
Source: MADLAB.CC
Source: Stelarc.org
