1 Technology and Commercial Innovation in Agri-TechCapacity Building and Networking Project To Support the UK-Turkey Partnership in the Agri-Tech Sector 6 March 2017, Albert Long Hall, BOĞAZİÇİ University, Turkey Prof Bruce Grieve, e-Agri Sensors Centre Electrical & Electronic Engineer, University of Manchester (UK)

2 Traditional View of Agriculture?

3 Why are Sensors & ICT important to Agri-Food?Ubiquitous! Legislative & Regulatory Market Forces Technology City Planning

4 Integrated Reality: No Single Solutionsmart sensing & monitoring BIG DATA smart control smart analysis & planning Image courtesy of S. Wolfert

5 Agri-Tech Involves the Whole Supply Chain and BeyondImage courtesy of S. Wolfert Smart Farming tracking/& tracing Smart Logistics Domestic IoT Fitness / Wellbeing Health Source: Hisense.com 5

6 New Business Models based on Big DataBasic data sales commercial equivalent of open data (e.g. FarmMobile) Product innovation use data to improve your product (machinery industry, e.g. John Deere, Lely’s milking robots) Commodity swap data for data (e.g. between farmers and (food) processors to increase service component) Value chain integration use data to control the whole chain (e.g. Monsanto’s Fieldscript) Value net creation pool data from the same consumer (e.g. AgriPlace) See: Arent van 't Spijker: "The New Oil - using innovative business models to turn data into profit“, 2014 Image courtesy of S. Wolfert

7 Redefining Industry Boundaries (1/2)(according to Porter and Heppelmann, Harvard Business Review, 2014) 3. Smart, connected product + + 2. Smart Product 1. Product Image courtesy of S. Wolfert

8 Redefining Industry Boundaries (2/2)farm management system farm equipment system weather data system irrigation system seed optimizing system field sensors irrigation nodes application seed optimization performance database weather data weather forecasts weather maps rain, humidity, temperature sensors 5. System of systems Your company planters farm equipment system combine harvesters tillers 4. Product system Image courtesy of S. Wolfert 8

9 Restructuring of Agri-Tech World to Address New Data RealityAgBusiness Monsanto Cargill Syngenta ... Farm Data Start-ups Tech Start-ups Farm Farming Cooperatives Open Ag Data Alliance ... Data Start-ups Farm Venture Capital Anterra Founders Fund Kleiner Perkins ... Tech Start-ups Farm Tech Companies Google IBM Intel ... Ag Tech John Deere Trimble Precision planting ... Farm Farm Image courtesy of S. Wolfert

10 The e-Agri concept is unique as it seeks to inform the electronics & IT community of the distinctive needs of modern agronomy & food science… ... So that they can fundamentally engineer new systems and “e-” devices for reducing waste, increasing yields and improving nutrition. Photos courtesy of Syngenta

11 e-Agri Food Network+ STFC Stakeholders = exploitation ICT, energy& electronics providers arable farmers Science & Tech R&D = technology push seed breeders retailers mechanical eng materials eng crop protection businesses distributors comp sci chem eng consumer products processors e-Agri = systems integration chemistry biology maths furtigation businesses plant sci physics fuel processors Brooks World Poverty Institute livestock Business School textile processors Agronomy Livestock Soils Food Tech Ext Res dairy Sustainable Consumption Institute food processors grain processors Business & Social R&D = market pull Agrimetrics CHAP CIEL Agri-EPI Innovate-UK government consumers regulators levy bodies NGOs £16M HEFCE Partnership

12 TRL Positioning of e-Agri CentreDe-risked for licensing or spin-out Take ideas to field trials & early business models = Stage Gate TRLs Not ‘blue skies’ – some basic concepts developed Business or Technology inspired idea for sustainable agriculture / food supply

13 Partnership Structure = Technology & Commercial InnovationTechnology delivery Systems Integrator (SME or applied academic) Structure all elements of the new technology supply chain to meet the product delivery (mostly SME) Distribution & Maintenance Service Providers (SME) Commercial pull through Commercial partner prepared to sell product offer (SME or Large) End User prepared to test product, in return for early adoption (Large)

14 “CHEAP & CHEERFUL” (Prof Lutz Plümer, University of Bonn, Germany) Smart, Ubiquitous, Agri-Electronics but Dominated by Theme 1

15 e-Agri: Some Examples …Sensors for protecting crop YIELD A sensor-network of 24/7 in-field disease pressure monitors Sensors “trick” pathogen into germination on mimic surface £2.5M Leveraged funding, TSB Crop Protection competition Sub-Soil Imaging New visualisation tool for seeds breeders Tracks how efficiently the root bundles are in drawing upon the water and nutrients in the soil. An in-field tool for early isolation and delivery of tomorrow's climate tolerant food crops Plastic Electronic-sensor Tagging (PET) UK, 4.1M tonnes of food that could have been eaten is thrown away A technology platform for recording produce temperature stress profiles from low cost, battery-free, plastic RFIDs £1.5M Leveraged funding from TSB Plastic Electronics competition

16 e-Agri: … and a Few More. iMAGiMAT™: SARIC Proposal on livestock protein conversion and gait monitoring – linked to biometric analysis of pigs. AB-Agri, Bristol Robotics Lab, SRUC Pollinator and Insect Pest Scanning: Optical back scatter analysis of iridescence of insect wings – ‘N2’ concept with Lancaster (c/o Andrew Wilby) Saturn-Sense: MIP sensors for P measurement (+ N, Ca, S, Mg, etc.) within hydroponics

17 Breaking the Barriers from ‘Farm to Fork’ “CHEAP & CHEERFUL” (Prof Lutz Plümer, University of Bonn, Germany) Low Cost, Close Proximity, Hyperspectral Sensors

18 UoB Hyperspectral Camera SystemLinear scanner

19 What?: Hypercube ImagesSource: Principles of Hyperspectral Imaging Technology, Elsevier, 2010

20 Exploiting Plant’s Natural Sunscreen1.16 * (R800-R670) (R800+R )

21 Early Prototype Light BoardContains over 1000 LEDs Designed to measure 530nm, 570nm, 670nm, 735nm, 830nm LED spacing designed to give uniform illumination over field of View Bright light source enabling faster image acquisition

22 Cercospora leaf spot on Sugarbeet𝐶𝐿𝑆= 𝑅 698 − 𝑅 𝑅 𝑅 570 − 𝑅 734 RGB CLS

23 The ‘Alpha’ Handheld Unit20 Discreet Wavelengths Hand-held Wi-Fi Connectivity Battery Powered Automatic Imaging Sequence Wavelength (nm) Intensity (arb. Units)

24 Texture Profiling Comes Free?HSI Affected by Lighting Angle Make that an advantage – ‘2.5D Imaging’ Image courtesy of L. Plümer , (Univ Bonn) Image courtesy of I Hales, (Univ West England)

25 Fluorescence Imaging Comes Free?PSII: Photochemical & NPQ Imaging RGB / NIR Sensor Response LED Response Image courtesy of J. Scholes, (Univ Sheffield)

26 The “Tricorder” for < US$100Crop disease sensing Early weed / plant detection; Speciating insect disease-vectors; Carbon footprint (protein in crops); Leaf viral symptoms; Harvested fruit bruising & sugars Soil organic carbon Aquatic parasites Frogs in salads ... GPS Camera Micro Processor Web linked + Subtle light manipulation = Active Close-Proximity Hyperspectral Imaging Fluorescence Imaging Stereo photometric Imaging Solid State Active Ellipsometry Image courtesy of CBS Inc.

27 Cercospora & Kayseriseker Sugar Turkey – ‘A brick wall’ ???Sugar beet pricing is made according to sugar percentage in sugar beet. As we know from the several researches, Cercospora is causing the sugar ratio in the sugar beet to fall. For instance, if the sugar percentage of sugar beet is %18, Cercospora would cause the reduction of the sugar percentage to %16. As a result of this reduction, for 1.5-million-ton sugar beet there will be ton sugar loss and this situation leads to an annual $ loss for our factory. Our aim is overcome this drawback by detecting the infected sugar beets as soon as possible and making a profit instead of the mentioned money loss. Şeyma Çakmak

28 Syngenta Plant Speciation: Hyperweeding - Weed ControlWeed control is becoming increasingly difficult due to herbicide resistant weeds and restriction of herbicides due to higher regulatory demands Resulting problems: In cereals, resistant blackgrass a severe problem with no good solution. Other examples worldwide Weed control in minor crops, such as vegetables, now extremely problematic as older herbicides have been de-registered In general active ingredients being rate restricted at sub optimal rates There is an urgent need to examine alternative or complimentary technologies for weed control to allow growers to grow crops profitably Directed treatment is one such technology

29 On-Tractor / Field Robot System: Plant / Weed SpeciationDirection of travel Tractor or “robot” Plants illuminated with specific wavelengths Reflected frequencies define plant type Computer decides what to treat Nozzle or laser is directed to targets weeds

30 On-Tractor / Field Robot System: Plant / Weed Speciation

31 Spin-out: to Protein Control

32 Spin-out to Smallholders FarmersHandy Hyperspectra: Crop stress sensing on your phone 120M Smallholder farmers in India (480M worldwide) 1 in 10 have smartphones and full internet connectivity Add a cheap attachment to turn phone into hyperspectral imager = 12M possible customers now for agricultural management packages!

33 Where Next – Going Underground?Chemical plant research… …translated to plant breeding

34 Electrical Impedance Tomography BackgroundWhat is EIT?

35 Towards Subsoil Biotic PhenotypingClubroot (Plasmodiophora brassicae) is an important pathogen of Brassica crops including oil seed rape (OSR), both in the UK and worldwide. Up to 10% of cultivated land worldwide is infected by clubroot and it is found in all UK regions where OSR is grown. Identification of quantitative resistance requires quantitative measures of plant responses to the pathogen which is particularly problematical for below-ground diseases. Currently using phenomics approaches to characterise above- ground responses to this important disease (thermography - IBERS). A.    Water-related measurements Clubroot infection occurs via root hair penetration (primary infection) and root cortex invasion (secondary infection), and leads into typical above- and below-ground disease symptoms. Infected plants develop large galls in the root system whilst above-ground symptoms include wilting, stunting, chlorosis, premature senescence and, in severe cases, death. All symptoms are caused through gradual changes in primary and secondary host metabolism, alterations in cambial stem cell maintenance and differentiation, and perturbations of vascular development with reduction in xylogenesis. The latter has an impact on the plants’ ability of water uptake from root to shoot. Therefore, the water content and water use of inoculated plants seem to be a promising target to detect early clubroot infection in susceptible and intermediate plant sets. Imaging techniques and water use measurements indicated infection signals previously, so we would like to keep focussing on water-related changes during our phenomics measurements. (1)     Thermal Imaging: Plants which are transpiring are cooled due to evaporation. Clubroot induced xylogenesis reduction and associated water uptake problems from root to shoot changes water content. Stomata closure is the plants strategy for water loss avoidance, and has an impact on evapotranspiration. Reduced transpiration rates and the absence of cooling effects cause elevated leaf temperatures, which can be measured via thermal imaging. It is hypothesised that leafs of clubroot infected plants show temperature differences compared to healthy plants. In the previous experiment conducted at IBERS, plants are placed in front of a heated background and data images are taken for each individual. From each data image 25 spot measurements can be taken manually with ImageJ software to calculate the average temperature of these values for each plant. Due to changing greenhouse temperatures, the average temperature of clubroot-inoculated plants is compared with un-inoculated control plants. Elevated leaf temperatures were seen in both inoculated Winter Oilseed Rape varieties and Chinese Cabbage Wong Bok, but not in both Kale varieties. Reduced transpiration in Winter Oilseed Rape Cracker with a Mendel-based clubroot resistance was surprising, and could be due to stomata closure as a stress/ immunity reaction against P. Brassicae instead of a sign for clubroot infection. Since manual data analysis based on 25 spot measurements is quite time consuming, either the number of conducted thermal imaging measurements must be reduced to a minimum of twice per week, or a quick and more efficient data analysis method must be found. Currently, a chess board overlay to connect RGB- and Thermal Images, to cut out plant material from unwanted background for calculation of average plant temperatures is discussed. A successful application could result in thermal imaging measurements on a daily basis.

36 ElT for Clubroot Infection DetectionEIT to directly measure below-ground responses as a complementary method of identifying quantitative resistance to clubroot infection in Brassica crops. Infection leads to an inhibition of lateral root formation, repression of xylogenesis, a localised induction of vascular cambium activity leading to gall formation and altered water relationships, all of which can potentially be measured and quantified using EIT. Cumulative water uptake measured on a daily basis in control and plants inoculated with low and high spore concentrations. Control and P. brassicae-infected Chinese var Wong Bok 51 days post inoculation

37 ElT Experimental Set Up (Aug-Sep 16)Subject Brassica plants: Winter oilseed rape 3 Control Plants 3 Infected Plants (Plasmodiophora Brassica) Equipment 6 EIT Vessels (32 electrodes per cell) Low-cost Tomography System 2 (LCT2) HP 4192a Impedance Analyser

38 Volume Conductivity (Mean Values)Baseline correction by applying 3rd order polynomial trend-line to each set

39 Rate of Change of ConductivityfRate of Change of Conductivity (S / ms) ANOVA tests were performed using this data. The results, with 95% confidence, indicate that EIT can detect clubroot

40 Reference ObservationsPlant H2 Plant I3 After experiment, the plants were removed and soil washed from roots. These plants were later introduced inside an EIT vessel filled with water, to observe the conductivity surrounding the roots. The roots from the infected plant showed a lower conductivity.

41 Next Generation LCT4 ModuleLow cost (c.£200) Miniature IoT compatible Faster measurements Translatable to field applications Amenable to Electrical Impedance Spectroscopy

42 Simultaneous Above & Below Ground Sensing: c.$500 Intelligent Shelf?Below Ground – Electrical Impedance Tomography: Current Source & Voltmeter + MUX + Software = $ Above Ground – Active HSI: LEDs + Digital Camera + Software = $

43 Conclusions Agri-Food chains become more technology / data-drivenCan cause major shifts in roles and power relations among different players in agri-food chain networks Infrastructure and software development are key issues Significant socio-economic impacts; two scenarios: Strong integrated supply chain farmer becomes franchiser/contractor limited freedom in doing business Open collaboration network Farmer empowered through easier switch between suppliers Options for direct sales to consumers F F Image courtesy of S. Wolfert

44 Agri-Tech Adoption Needs Commercial & Technology InnovationNew Disruptive Technologies need new: methods of identify real not perceived needs business models and supply chain interrelationships UK Global Food Security Programme: “IKnowFood” (Liverpool, York & Manchester Universities) Launch Oct 2016, 4 years, €3.7M An opportunity to address the issues