vtt technical research centre of finland … · vtt technical research centre of finland ltd new...
TRANSCRIPT
VTT TECHNICAL RESEARCH CENTRE OF FINLAND LTD
New sensor technologies for drone agriculture applications
Tehoa maatalouteen droneilla, 21.3.2016, SaloHeikki Saari, VTT Microspectrometers
218/03/2016 2
ContentsWhy new drone sensors are needed in agriculture? State-of-the-art of drone assisted crop precision farming –
Example Sequoia Drone imaging system Novel drone multi- and hyperspectral imaging systems Hyperspectral iPhone demo Cubic inch size MEMS-FPI based hyperspectral imager with
mobile camera module Drone multispectral imager with 12 channels based on four RGB
cameras. RFID sensors with drone used for the remote readoutWhat can be done to create Finnish drone system products for
agriculture? Conclusions
318/03/2016 3
Why new drone sensors are needed in agriculture? Climate change. How to prevent crop losses in rapidly changing growth
conditions - new plant diseases, unprecedented weather conditions. Human chemical exposure. How grow effectively with drastically reduced use
of fertilizers and pesticides (natural and organic food). In general there is a need for integration of existing Precision Farming (PF)
equipment, sensor modules, Digital Support System (DSS). There is a need for improved DSS for variable-rate-technology fertilization including
sensor data AND economic costs for Drone based PF approaches. There has been lots of research efforts on drone based sensors that has been
shown to provide useful information on nutritional status, water stress and plant health. The biggest problem of existing solutions is the cost of advanced drone sensors
used in research for example hyper- and multispectral imagers. Focus of this presentation is to provide a short introduction what new
technologies there are for cost effective multi- and hyperspectral agriculture drone sensors
418/03/2016 4
Sequoia 4 channel multispectral sensor for drone agriculture applications
Ref. http://www.parrot.com/usa/companie
s/sequoia/Ref.https://www.pix4d.com/wp-content/uploads/2016/02/Sequoia_Flyer.pdf
518/03/2016 5
Sequoia 4 channel multispectral sensor for droneagriculture applications
Ref. https://www.pix4d.com/wp-content/uploads/2016/02/Sequoia_Flyer.pdf
• In Sequoia systemthere are four camerasproviding images at 4 spectral bands in Green, Red, Red-Edgeandd VNIR
• The selected fourspectral bands provideuseful information butfor more accuratedetection of plantdiseases, watershortage, nitrogen and biomass content morespectral bands arerequired.
718/03/2016 7
Novel drone multi- and hyperspectralimaging systems
MEMS-FPI based hyperspectral cameras based on smart phonecamera modules (ref. Rissanen et.al. ”MEMS FPI-basedsmartphone hyperspectral imager”, SPIE 9855 to be published).MEMS-FPI based hyperspectral imager with mobile phone
camera module in cubic inch size.Multispectral cameras based on multi passband filters and RGB
image sensors.
818/03/2016 8
Hyperspectral iPhone demo
Tunable MEMS FPI chipsWavelength ranges:
• 450 – 550 nm/ single chip• 450 - 700 nm/ tandem
• 800 – 1050 nm
Mobile application potential:• Counterfeit identification
• Authentication• Vis/NIR food sensing
• Heath/Wellness
Future for ubiquitous sensing • Software programmable• Cloud-based data• Novel applications through
mobile app developers
MEMS Diagnostics Environment
918/03/2016 9
MEMS-FPI based hyperspectral imager with mobile phone camera module
1 Mpix hyperspectral imagerconcept
One MEMS-FPI between mobile camera lenses
FOV defined by the hyperspectralimager lens
Size of the module would bearound cubic inch (25 x 25 25 mm)
1018/03/2016 10
VTT multispectral imager based on multipassband filters and RGB camera
400 500 600 700 800 900 10000
0.2
0.4
0.6
0.8
1
Wavekength/[nm]
Tran
smis
sion
600 650 700 750 800 850 9000
0.04
0.08
0.12
0.16
0.2
Multi passband filter and R pixelsMulti passband filter and G pixelsMulti passband filter and B pixels
Wavekength/[nm]
Rel
ativ
e se
nsiti
vity
400 500 600 700 800 900 10000
0.2
0.4
0.6
0.8
1
Red/R pixelsGreen/G pixelsBlue/B pixels
Wavelength/[nm]
QE
3 Mpix multispectralimager concept
Multi passband filteroptimised for Red Edge measurement.
The Multi passbandfilter can beoptimised for anyrange inside 400 –1000 nm.
Red edge multi passbandfilter transmission
Quantum efficiencies of R-, G- and B-pixels
Combined sensitivities of multi passband filter and of R-, G- and B-pixels
1118/03/2016 11
VTT multispectral imager based on multi passbandfilters and RGB camera Multispectral imager for three spectral bands can be built using multi passband filter and
long&short pass filters with RGB camera. Optics can be built based on three low cost microvideo objectives.
1218/03/2016 12
Concept for a 12 band multispectral imager basedfour diffrent multi passband filter camera modules It is possible to build a 12 channel multispectral camera based on four 3 cahnnel multispectral
cameras. The size of the system with a single board PC is around 65 mm x 125 mm x 125 mm and the
weight is < 750 g
1318/03/2016 13
RFID sensors with drone used for remotereadout
There are new sensor technologiesthat can be used for the developmentof sensors with RFID type readoutfrom drone. A new vision for smart objects and the
Internet of Things wherein drones interact with wirelessly-powered, long-range, ultra-high frequency radio frequency identification (UHF RFID) tags outfitted with sensing capabilities. http://arxiv.org/pdf/1507.02373.pdf The challenge is how to develop
disposable sensors with dronereadout
1418/03/2016 14
RFID sensors with drone used for remotereadout In a paper recently posted
on arXiv, a team of researchers has proposeda hybrid approach usinglong-range UHF RFID sensors that are dirt cheapand require no powersource, combined with a adrone that can talk to them. Is this feasible? Ref.
http://diydrones.com/profiles/blogs/adding-uhf-rfid-sensors-to-iris-for-aerial-farm-management
1518/03/2016 15
What can be done to create Finnish dronesystem products for agriculture?
The whole value chain from a farmer to sensor module producermust be involved in the definition economic constraints i.e. what arethe benefits versus cost of a drone based service. How to organisethis? The drone sensor market is global but to enter the market there is a
need for national piloting – How can we get this done in Finland? Is it possible to define the most promising Finnish agriculture
applications of drone systems? What sensoring is required for these applications? Is it possible to develop disposable sensors that can be read fro
drones for soil parameters and/or plant diseases?
1618/03/2016 16
Conclusions
There are lots of new sensor technologies that can be used in drone based precision farming. VTT has solutions for cost effective hyper- and multispectral
drone imaging systems that could developed as tools for droneassited agriculture. An other promising sensor approach is RFID sensors with drone
used for remote readout.
[email protected], +358 40 5891254