Page 57 - EE Times Europe Magazine | June2020

EE|Times EUROPE — Sensors Insert 57

Microfluidic Technologies Diagnose World’s Return to Normal

footprint. “A microfluidic cartridge is dedi- mid-sized companies. Market consolidation to develop microfluidic tests. Bosch Health-
cated to one type of test, and different types has accelerated to the extent that a group of care, for instance, has teamed with Randox
of tests need to be developed,” said Clerc. about 15 players now accounts for more than Biosciences and R-Biopharm to develop tests
“A company can develop two to three tests 75% of the market. “Large companies do not to implement on its Vivalytic platform.
per year, but there are hundreds to develop.” take the risk of technical development and “If we look ahead two or three years and
That’s where collaborations and acquisitions prefer to invest in already-developed technol- if we reach a dozen partnerships, that’s
make sense. ogies,” said Clerc. potentially two or three additional tests per
At the same time, an acquisition by a larger partnership per year,” said Clerc. “This is
FOSTERING RELATIONSHIPS company can provide startups with an estab- starting to provide an extended range of tests
In the health-care industry, barriers on the lished distribution network and improved for users.” ■
road to success are high, and acquisitions are logistics support. IQuum (Roche), BioFire
often the best solution to remain compet- Diagnostics (bioMérieux), and STAT-Dx Anne-Françoise Pelé is editor-in-chief of
itive. In the past decade, large diagnostics (Qiagen) are good examples of microfluidic eetimes.eu. Maurizio Di Paolo Emilio is a
companies such as bioMérieux, Roche, and technology developers whose approaches took staff correspondent at AspenCore, editor of
Qiagen have acquired promising microfluidic off once the companies were acquired. Power Electronics News, and editor-in-chief
technologies through the purchase of small or But collaboration remains the fastest way of EEWeb.

SPECIAL REPORT: NEUROMORPHIC COMPUTING
The Slow but Steady Rise of the Event Camera

By Tobi Delbrück

oughly a billion dollars of investment in CMOS image sensors
(CIS) over the past 20 years has led to the current market,
where these beautiful imagers are produced by the billions each
Ryear. As CIS became a commodity, neuromorphic silicon retina
“event camera” development languished, gaining industrial traction
only recently, when Samsung and Sony put their state-of-the-art image
sensor process technologies on the market.
Our event camera, introduced at the 2006 International Solid-State
Circuits Conference (ISSCC), included huge, 40-µm pixels using a
350-nm process. Even then, CIS pixels were down by about a few
microns. In 2017, Samsung published an ISSCC paper on a 9-µm-pixel,
back-illuminated VGA dynamic vision sensor (DVS) using their 90-nm
CIS fab. Meanwhile, Insightness announced a clever dual-intensity +
DVS pixel measuring a mere 7.2 µm. assisted me in building the first USB DVS camera. We sold several
Both Samsung and Sony have built DVS solutions with pixels under hundred 128 × 128-pixel DVS cameras to neuromorphic community
5 µm based on stacked technologies in which the back-illuminated early adopters who were not ASIC developers. This pixel architecture
55-nm photosensor wafer is copper-bumped to a 28-nm readout wafer. is the foundation of all subsequent generations from all the major
Amazing increases in event readout speed have also resulted from players (even when they don’t say so on their websites). The DVS
industrial development. These clever designs are bringing DVS pixels brings a “unique selling proposition” over previous silicon retinas
down to the sizes of standard global-shutter machine-vision and auto- and standard cameras, owing to its sparse, quick spiking output that
motive camera pixel sizes. It means that DVS has a fighting chance to responds reliably to low-contrast natural scenes while offering great
establish itself as a viable mass-production vision-sensor technology in dynamic range and speed. Early DVS cameras allowed neuromorphic
the same “megapixel race” that has consumed CIS for decades. researchers to play with the technology to determine its potential.
The development of neuromorphic silicon retinas is a great example A decade later, conventional machine-vision and robotics research-
of faith meeting practical reality. The development of silicon retina ers did the same. This would not have happened without my students
event cameras goes back to 1989 with Kunihiko Fukushima’s Reticon Patrick Lichtsteiner, Raphael Berner, and Christian Brandli, who now
and the work of Carver Mead and Misha Mahowald at Caltech in the lead several startups. The other key was long-term support from UZH
early 1990s. and ETH for basic technology development and funding from the
I joined this effort as a graduate student at Caltech with Mahowald European Commission’s Future and Emerging Technologies initiative.
and Mead as mentors. We neuromorphic engineers believed we could
build a camera that worked like the biological eye. The reality after a GROWING ECOSYSTEM
IMAGE: SHUTTERSTOCK big (i.e., expensive) and too noisy (i.e., they made terrible pictures). startups such as Insightness (recently acquired by Sony), iniVation
decade of early work was that our “silicon retina” pixels were vastly too
Similar to what occurred with CMOS image sensors, event camera
(which carries on the iniLabs mission), Shanghai-based CelePixel,
Just as important, they didn’t offer sufficient advantage over CIS.
and well-heeled Prophesee are established, with real products to sell.
All this early development was taking place concurrently with
Others will surely follow.
the constant improvement of CIS. A breakthrough of sorts occurred
Recently, mainstream computer-vision researchers introduced to
during our work on the European project called CAVIAR, when Patrick
Lichtsteiner and I came up with the DVS pixel circuit. Anton Civit
event cameras (mainly via academic collaboration or through our
www.eetimes.eu | JUNE 2020