SmartFarm: Data Integration & Systems for Agriculture -based, …€¦ · SmartFarm: Data Integration & Systems for Agriculture -based, FEW Research. Chandra Krintz. Dept. of Computer

SmartFarm: Data Integration & Systemsfor Agriculture-based, FEW Research

Chandra KrintzDept. of Computer Science

UC Santa Barbara

Database Integration Workshop: Building the Data Capacity for Food- Energy-Water (FEW) Research - Sept. 11, 2018

Data Analytics: Making Sense of It All

Data Analytics: Making Sense of It All

Sister is a lawyerChildren play hockey

Dislikes snakes2013 Salesman of Year

Owns an RV

Hates doing dishes

SalesmanDislikes Actor Robert Redford Likes online news sites

Has 3 kids

Male

Owns a SmartTV

Age 38-40

Mother lives in Florida

Is politically active

Major life insurance holder

Works out at a gymLikes basketball Likes spicy food recipes

Household Income: 150000Reads crime dramas Likes Jimmy Fallon

Republican

Is active on TwitterDrives a Ram truck

Likes hiking House value: $500,000

Cloud + Data Analytics: Have Revolutionized Commerce

What will you buy?

When will you buy it?

What will you pay?

Inference and PredictionInternet Activity

Math andStatistics (Code!)

Cloud + Data Analytics: Have Revolutionized Commerce

What will you buy?

When will you buy it?

What will you pay?

Inference and PredictionInternet Activity

Math andStatistics (Code!)

What Else Can We Revolutionize With It?

Our Goal: Tailor Cloud & Data Analytics To Address the Critical Needs& Complex Challenges of Food Production

The world needs more food for a growing population.(increase yields sustainably)

Invasive pests and disease threaten production(detect, monitor, and predict spread)

We use 80% of fresh water for growing food. (precision application of inputs)

30% of global energy is used to produce food22% of greenhouse gases come from agriculture(integrate the FEW nexus into decision making)

We lose 30+% of food we produce to spoilage(data-driven harvest & delivery; end-to-end tracing)

Worker shortages and high labor costs(increase automation & operating

efficiencies)

Inference & Prediction

IrrigationScheduling

Disease/PestManagement

Tracking from Farm to Fork

Public Clouds

Public Clouds

Problems:• Moves vast amounts of data (to code/services)

• Unreliable, if available at all• Costly ($$, power)

• Cloud/Internet designed for reads not writes• Farmers lose control/ownership over their data• Underlying technology is constantly changing

Edge Clouds

Edge Clouds(similar to EPAdata appliancebut with analytics support)

Edge Tier Public Cloud TierDevices

Edge Clouds Public Clouds

Edge Tier Public Cloud TierDevices

Edge Clouds Public Clouds

Move Code (services) and/or

Data

Controlled, Anonymized Sharing

Edge Tier Public Cloud TierDevices

Edge Clouds Public Clouds

Regional Tier

Community and University

Clouds

SmartFarm Research

On-farm, autonomous and self-managing systemsSensing and analytics; ground and air vehicles

Very low-cost (potentially unreliable) sensingStatistical techniques (software) reconstitute information

Fused data analytics, access control, machine learning, & diagnostics

Problem fociSoil health and microclimate mapping & analysis

Management zone identification

Frost prediction and damage mitigation

Precision application of water, pesticide, fertilizer

Tracking from farm to fork (or landfill)

Key Challenges with Data Integration for Ag

CODE!

Solutions differ widelyLocal/regional vs global

Multi-scale: time and space

Data/solution consumers can beHuman: Verification, visualization, diagnostics required

Machine: amenable to diff. analysis techniques (mixed methods, comp. tools)

Collection: Cost, ease-of-use, robustness, & privacy-control ALL matter

Challenging data sources and sensors (+ lack of standards/sharing)Heterogenous, geographically distributed, proprietary

Must have the option of moving code (APIs) to data as well as data to code

Must leverage similar extant efforts & cross-sectional collaboration

A New Kind of Computer Science Research

• Problem driven and empirical• Food-Energy-Water nexus

• Societal and regional impact• Multidisciplinary collaboration• Repeatable, demonstrable, applied

(tech-transfer ready)• Engages students & the community

Thanks!

Collaborators: UCSB, LREC, CalPoly, Fresno State, Powwow Energy, Sedgwick Reserve, Private Growers

Support: Google, Huawei, IBM Research, Microsoft Research, NSF, NIH, California Energy Commission

[email protected], [email protected]://www.cs.ucsb.edu/~ckrintz/racelab.html

Students: