MIT develops indoor navigation for drones
MIT's MiFly drone indoor navigation system works with reflected radar waves. This makes it possible to determine the position indoors with an accuracy of 7 cm.
(Image: MIT)
Scientists at the Massachusetts Institute of Technology (MIT) and the University of Michigan have developed an indoor navigation system for drones that works without computer vision or lidar. Instead, the scientists based their MiFly system on the evaluation of reflected radio waves. The technology also works in dark environments.
Under the open sky, drone navigation is child's play. A drone's exact location can be determined using GPS. It is more difficult when drones are used in closed buildings, such as warehouses, and need to find their way around, as GPS does not work indoors due to weak signals. Technologies such as computer vision and lidar also have their pitfalls there. Computer vision fails in poor lighting conditions, while lidar has difficulties with smooth walls. In addition, orientation with lidar is difficult in environments with repetitive elements. This applies, for example, to long rows of shelves in warehouses that all look the same.
Dual radar
MIT's MiFly system works with two radar systems housed in a drone, as the scientists explain in the paper “6D Self-Localization of Drones using a Single Millimeter-Wave Backscatter Anchor” (PDF). The radar emits millimeter-wave signals. They can penetrate everyday materials such as cardboard, plastic, and interior walls and are also used in modern radar systems. The emitted signals are reflected by a backscatter tag, which can be placed anywhere in a room and requires very little power.
Empfohlener redaktioneller Inhalt
Mit Ihrer Zustimmung wird hier ein externes YouTube-Video (Google Ireland Limited) geladen.
Ich bin damit einverstanden, dass mir externe Inhalte angezeigt werden. Damit können personenbezogene Daten an Drittplattformen (Google Ireland Limited) übermittelt werden. Mehr dazu in unserer Datenschutzerklärung.
As the reflected radar signals can also originate from the surroundings, the tag adds another frequency to the reflected signal. The drone is thus able to distinguish the required signal reflections of the tag from those of the surroundings.
A single tag is not actually sufficient to determine the location of a drone. It can only be used to determine the distance between the drone and the tag. The researchers therefore use a trick. They mount one radar horizontally and the other vertically in the drone. The polarization is correspondingly horizontal and vertical. One radar sends the signals horizontally, the other vertically. The researchers took the polarization into account in the tag so that it is possible to isolate the signals sent by the two radars. The scientists applied different modulation frequencies to the vertical and horizontal signals to largely rule out interference.
Positioning
With the combined signals, it is possible to determine the spatial position of the drone. To navigate, however, it must estimate its position in space regarding its six degrees of freedom. In addition to values for forward, backward, left, right, up, down as well as pitch, yaw and roll, the flight paths are also considered.
“The rotation of the drone adds a lot of ambiguity to the millimeter wave estimates. That's a big problem because drones rotate quite a bit during flight,” says Laura Dodds, research assistant at MIT and co-author of the study.
Videos by heise
To take the drone's position in space into account and enable more accurate signal estimates, the scientists used the drone's internal position sensors, which also output altitude and acceleration in addition to position. The researchers combined this information with the reflected signals. MiFly can thus determine the drone's position in space within milliseconds.
MIT tested the MiFly drone in the laboratory, in the flight room and in dark tunnels under the university campus. The position of the drone could be determined to within 7 cm in all tests. This worked up to a distance of 6 m during the day.
The MIT researchers are certain that the distance can still be significantly increased, for example by using high-performance amplifiers or improving the radar and antenna design. The scientists also want to add autonomous navigation capabilities to MiFly. The drone could then independently determine a flight route based on radar reflections. However, it may be some time before this results in a commercially marketable system.
(olb)
