Research Better Orientation for Robots

Author / Editor: Simone Käfer / Alexander Stark

Turbotrack is designed to improve robot collaboration and accuracy. The system uses RFID tags to identify and locate objects in a three-dimensional space. The challenges are in the interaction of speed and accuracy.

Related Companies

Using RFID, robots are now able to locate moving objects in a three dimensional space. The system was developed by MIT researchers.
Using RFID, robots are now able to locate moving objects in a three dimensional space. The system was developed by MIT researchers.
(Source: MIT / CC BY-ND 3.0)

Most robots use sensors or cameras and image processing systems to orient themselves. Like human vision, capabilities are limited and objects in a complex environment are quickly overlooked. Researchers from MIT (Massachusetts Institute of Technology) want to change that. That's why they developed Turbotrack. The system uses RFID tags on objects. A read/write device sends a wireless signal that is reflected by the RFID tag and objects near it and reflected back to the reader. An algorithm scans all reflected signals to identify the response of the RFID tag. Final calculations then use the movement of the RFID tag to improve localization accuracy.

To validate the system, MIT researchers attached one RFID tag to a cap and another to a bottle. A robot arm positioned the cap and placed it on the bottle, which was held by a second robot arm. In another demonstration, the researchers followed RFID-supported nanodrones during docking, maneuvering and flying. In both cases, the system was as accurate and fast as conventional vision systems. So there's no need to change technology. But Turbotrack also works in applications where image processing fails. That's why, the researchers report, the high-frequency (RF) signals from RFID can identify targets even in complex environments and through walls. In the words of Fadel Adib, Assistant Professor and Principal Investigator at the MIT Media Lab and founding director of the Signal Kinetics Research Group: "When you use RF signals for tasks typically performed by vision systems, you're enabling robots to do not just human things, but superhuman things as well. And this is even financially scalable, because the costs for an RFID tag are in the cent range.

In manufacturing, the system could enable more precise and versatile robotic arms, for example for picking up, assembling and packing items along an assembly line.

Speed Combined with Precision

The challenge is to improve the coordination of speed and precision. To be precise, it can take a few seconds to find a moving object; if you increase the speed, the systems become less accurate. In order to achieve both precision and speed, the researchers were inspired by an imaging technique called "Super-Resolution Imaging". These systems combine images from different angles to create a much better image. "If something moves, you get more perspectives in the pursuit, so you can use the movement for accuracy," Adib explains the idea. Therefore, the system also sends broadband signals and the researchers have developed a "space-time superresolution algorithm". If the object moves with the RFID tag, the signal angle changes slightly. By constantly comparing the changing distance of the RFID signal with the distances of the other signals, the system can find the RFID tag in three-dimensional space. All this happens in fractions of a second.

This article was first published by MM Maschinenmarkt.

Original by Simone Käfer / Translation by Alexander Stark