The 3D camera designed for a flying robot can identify small objects measuring 15 by 20 cm from 7 m away. Image: Fraunhofer Institute |
Like
a well-rehearsed formation team, a flock of flying robots rises slowly
into the air with a loud buzzing noise. A good two dozen in number, they
perform an intricate dance in the sky above the seething hordes of
soccer fans. Rowdy hooligans have stormed the field and set off flares.
Fights are breaking out all over, smoke is hindering visibility, and
chaos is the order of the day. Only the swarm of flying drones can
maintain an overview of the situation.
These
unmanned aerial vehicles (UAVs) are a kind of mini-helicopter, with a
wingspan of around 2 m. They have a propeller on each of their two
variable-geometry side wings, which lends them rapid and precise
maneuverability. In operation over the playing field, their cameras and
sensors capture urgently-needed images and data, and transmit them to
the control center. Where are the most seriously injured people? What’s
the best way to separate the rival gangs? The information provided by
the drones allows the head of operations to make important decions more
quickly, while the robots form up to go about their business above the
arena autonomously—and without ever colliding with each other, or with
any other obstacles.
A
CMOS sensor developed by researchers at the Fraunhofer Institute for
Microelectronic Circuits and Systems IMS in Duisburg lies at the heart
of the anti-collision technology.
“The
sensor can measure three-dimensional distances very efficiently,” says
Werner Brockherde, head of the development department. Just as in a
black and white camera, every pixel on the sensor is given a gray value.
“But
on top of that,” he explains, “each pixel is also assigned a distance
value.” This enables the drones to accurately determine their position
in relation to other objects around them.
Sensor has a higher resolution than radar
The
distance sensor developed by the IMS offers significant advantages over
radar, which measures distances using reflected echoes.
“The
sensor has a much higher local resolution,” says Brockherde. “Given the
near-field operating conditions, radar images would be far too coarse.”
The flying robots are capable of identifying even small objects
measuring 20 by 15 cm at ranges of up to 7.5 m. Moreover, this distance
information is then transmitted at the very impressive rate of 12 images
per second.
Even
when there is interfering light, for example when a drone is flying
directly into the sun, the sensor will deliver accurate images. It
operates according to the time-of-flight (TOF) process, whereby light
sources emit short pulses that are reflected by objects and bounced back
to the sensor. In order to prevent over-bright ambient light from
masking the signal, the electronic shutter only opens for a few
nanoseconds. In addition, the sensor also takes differential
measurements, in which the first image is captured using ambient light
only, a second is taken using the light pulse as well, and the
difference between the two determines the required output signal. “All
of this happens in real time,” adds Brockherde.
The
3D distance sensors are built into cameras manufactured by TriDiCam, a
spin-off company of Fraunhofer IMS. Jochen Noell, TriDiCam’s managing
director, admits: “This research project has presented us with new
challenges as regards ambient operating conditions and the safety of the
sensor technology.” The work falls under the AVIGLE project, one of the
winners of the ‘Hightech.NRW’ cutting-edge technology competition which
receives funding from both the Land of North Rhine-Westphalia and the
EU. The IMS engineers will be presenting their sensor technology at the
Fraunhofer CMOS Imaging Workshop in Duisburg on June 12 and 13 this
year.
Conducting
intelligent aerial surveillance of major events is not the only
intended use for flying robots. They could also be of benefit to
disaster relief workers, and likewise to urban planners, who could
utilize them to produce detailed 3D models of streets or to inspect
roofs in order to establish their suitability for solar installations.
Whether deployed to create virtual maps of difficult-to-access areas, to
monitor construction sites or to measure contamination at nuclear power
plants, these mini UAVs could potentially be used in a wide range of
applications, obviating the need for expensive aerial photography and/or
satellite imaging.
Source: Fraunhofer Institute