New 3D sensor should help UAVs avoid fender-benders
By Ben Coxworth
May 10, 2012
Hovering unmanned aerial vehicles (UAVs) – most of which take the form of quadrocopters – are currently being developed for a wide range of applications. Some of these include the delivery of supplies to remote locations, urban reconnaissance, and military operations. Whether they’re flying solo or in organized swarms, however, they constantly need to be aware of potential collision hazards, both mobile and stationary. While various technologies are already being utilized for this purpose, Germany’s Fraunhofer Institute for Microelectronic Circuits and Systems has developed a new 3D CMOS sensor, that promises particularly good performance.
The sensor is based around the time-of-flight process, in which the relative distances of objects are determined by shooting short bursts of light at them, and measuring how long it takes that light to reflect back to an onboard receptor. In order for the reflected light pulses not to be masked by bright ambient light, the receptor’s shutter only stays open for a few nanoseconds. The sensor processes the data at a rate of 12 images per second.
Every pixel of these images is assigned a gray value, along with a distance value. That level of differentiation allows the system to identify objects as small as 20 x 15 centimeters (8 x 6 inches), from distances of up to 7.5 meters (24.6 feet).
Based on this constant stream of data, the UAV can determine its location in space, relative to the objects surrounding it.
While the process may sound somewhat similar to radar or sonar, its higher resolution apparently makes it much better suited to the close-proximity flying that quadrocopters are frequently required to do.
The 3D sensors have been built into cameras made by a Fraunhofer spin-off company, TriDiCam.