Teaching robots to copy human movement
Having two arms doesn't make you a juggler. The same principle applies in robotics where even the most dextrous of bots must be programmed to move according to a particular task. Input systems based on laser tracking are used in industrial robotics to achieve this, but Fraunhofer researchers are looking to streamline the process significantly with a device that uses inertial sensors to track movements in free space. In other words, you can teach a robot new tricks just by showing it the required action.
The key to the system is its ability to analyze how the sensors on the input device interact.
"We have developed special algorithms that fuse the data of individual sensors and identify a pattern of movement," says project leader Bernhard Kleiner of the Fraunhofer Institute for Manufacturing Engineering and Automation IPA in Stuttgart. "That means we can detect movements in free space."
The result is billed as a system that simpler and less labor intensive than laser tracking or camera-based approaches and researchers see the potential applications going beyond industrial robotic arms – for example it could be used for gait analysis and to help control the motors in an active prostheses.
"With the inertial sensor system, gait analysis can be performed without a frame of reference and with no need for a complex camera system," says Kleiner.
Researchers are also working on combining the system with EMG sensors that detect muscle movement to improve regulation of prosthetic limbs.
"While standard EMG sensors consist of individual electrodes that have to be positioned precisely on the muscle, our system is made up of many small electrodes that attach to a surface area. This enables us to sense muscle movements much more reliably," says Kleiner.
The system will be on display at the Sensor+Test trade fair in Nuremberg from June 7 – 9 where visitors will see just how challenging it is to show a robotic arm (complete with catcher's glove) how to to catch a ball.