Robotics & Control in Agriculture Abubakr Muhammad Director, - - PowerPoint PPT Presentation

Robotics & Control in Agriculture

• EE361. Lectures on Control Engineering in Environment & Sustainability

Spring 2015, LUMS

Abubakr Muhammad

Director, Laboratory for Cyber Physical Networks and Systems Dept of Electrical Engineering SBA School of Science & Engineering Lahore University of Management Sciences (LUMS), Pakistan

Objectives

• PLO-7: Environment and Sustainability (PEC )
• Deeper motivation: Connecting technology to real-

world and societal grand challenges 

• Accessible introduction to cutting-edge research 
• Pay attention to the Right Problems! 
• Demonstrate how student involvement helps develop

high impact research 

Outline

• Motivation: The importance and context of

Agriculture in Pakistan

• Precision Agriculture – A systems perspective
• Feedback Control in Precision Ag

– Auto-steering – Variable rate control

• Ag Robotics: A new frontier in farm automation
• The case for Automation & Robotics in Pakistan
• Conclusions and outlook

Agriculture at the Center

Agriculture Industry Commerce

Agriculture in Pakistan provides

• Food
• Raw material
• Foreign trade

25% GDP, >50% of population in agriculture Pakistan’s Ag Profile Traditional: Rice, Cotton, Wheat, Sugarcane, Maize Upcoming: Livestock, Fisheries, Forestry, Horticulture

Agricultural Footprint

~25% under cultivation of which 80% is irrigated.

The Green Revolution in the Indus Basin

• Irrigation canal networks (pre- and post-partition)
• Emergence of agricultural research resulting in High

Yield Varieties HYV (1950s)

• New “miracle” seeds, fertilizers, mechanization,

groundwater pumping, multiple cropping, processing, storage, financial measures etc. (1960s-1980s)

• The boat we probably missed was the Gene

revolution! (1990s-) (GM crops: herbicide-, disease-, drought-, insect- resistant varieties)

Problems Related to Low Productivity

Poor Economics 1. Under-utilization or over-exploitation of cultivable land, manpower 2. Uneconomic holdings (farm sizes) and defective land tenure system 3. Pricing and subsidies Poor Methodologies 1. Insufficient or inefficient use of Inputs (pesticide, fertilizers, seeds, mechanization, irrigation) 2. Water issues: logging, salinity, scarcity 3. Agricultural research and extension / education Poor Infrastructures 1. Rural infrastructures and services (roads, energy, distribution, storage…) 2. Markets and financial institutions (access, credit, insurance, smuggling) Negative Driving Forces 1. Climate change, droughts, floods and natural disasters 2. Diseases and pests 3. Population growth and demographic transitions 4. Urbanization, globalization

Problems Related to Low Productivity

Poor Economics 1. Under-utilization or over-exploitation of cultivable land, manpower 2. Uneconomic holdings (farm sizes) and defective land tenure system 3. Pricing and subsidies Poor Methodologies 1. Insufficient or inefficient use of Inputs (pesticide, fertilizers, seeds, mechanization, irrigation) 2. Water issues: logging, salinity, scarcity 3. Agricultural research and extension / education Poor Infrastructures 1. Rural infrastructures and services (roads, energy, distribution, storage…) 2. Markets and financial institutions (access, credit, insurance, smuggling) Negative Driving Forces 1. Climate change, droughts, floods and natural disasters 2. Diseases and pests 3. Population growth and demographic transitions 4. Urbanization, globalization Robotics, Automation & Control

The Precision Ag Revolution

Measure and respond against variability while

• ptimizing returns.
• Courtesy. Tristan Perez, QUT, Australia

The Precision Ag Revolution

Measure and respond against variability while

• ptimizing returns.

Key technologies (1990-2010)

• Variable rate input
• GPS enabled auto-steering
• Satellite imagery
• Minimal / No tilling

Control Architecture

• ECUs on implements connected via CANBus
• Courtesy. Crop Protection, Australia

GPS enabled Auto-steer

Inter-row sowing

• Courtesy. Crop Protection, Australia

Using Satellite Imagery for Yield Maps

• Courtesy. Andrew Robson, UNE

Variable-Rate Treatment

• Once mapped, how to act? (see e.g. weed map

below)

• Variable-rate (local measurements) Vs. Fixed rate

(bulk measurement)

• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

Variable-Rate Treatment

• Real-time adaptive field spraying over weeds
• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

Automatic Sensing & Spraying

• Courtesy. Horticulture Innovation Australia, Sugar Research Australia

Active Spray Boom Suspension

• Active suspension systems to counter soil unevenness.
• The two hydraulic actuators counteract tractor yawing and

jolting by moving the sledge in the opposite direction.

• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

Modeling and Control

• Textbook quarter-car

suspension model

• Disturbance, control
• Disturbance step response

Control Specification (Freq. Domain)

• Tractor accelerations below 0.5 Hz are due to
• perator maneuvers
• Only vibrational modes of the boom below 10 Hz

contribute to an uneven spray deposition pattern .

• Therefore isolator should attenuate boom

accelerations between 0.5 and 10 Hz.

• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

Model Identification

• Plant identification
• Separate modes

– Rotational – Translational

Controller Performance in Field

• Sensitivity function

E(s) = S(s)R(s) – S(s)G(s)W(s) + T(s)V(s)

• Loop shaping
• Boom tip movement (with and

without control)

• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

Fertilizer Spreader

• Precise spreading of liquid menure
• Disturbance: Variable vehicle speed
• Variability: Slurry setpoint variation
• Ref. De Baerdemaeker et al. IEEE Control Systems Magazine, 2001

End of Part 1

Agricultural Robotics

• Courtesy. Eldert Van Henten, Waganengin University

Ag Robotics Platforms

• Farm mapping and autonomy
• Yield estimation e.g. Almonds and Apples
• Tree database e.g. Almonds
• Precision weed sensing
• Horticulture
• Courtesy. Horticulture Innovation Australia. CMU. USDA. Usyd, Bulent Ecevit Univ, Technion, Aalto

ACFR Platform

Robotics in Horticulture

• Agronomic solutions - crop

nutrition, canopy structure, pest numbers/identification, weed detection/removal, yield

• Physiological solutions -

flowering, fruit set, maturity indices (colour/sugar), forcasting/yield, abiotic stress (cold injury, drought, heat, salinity and metals)

• Social solutions – safe, skilled

and increased capability

• Ref. Anthony Kachenko, Horticulture Innovation Australia

Yield Mapping

• 3D vision algorithms
• Active and Passive Sensing
• Courtesy. USDA, ACFR

Fruit Picking Automation

• Courtesy. UC Davis

Automatic weed spot spraying

• Courtesy. Horticulture Innovation Australia, Sugar Research Australia

Automated Harvesting: Indoor / Greenhouse

• Courtesy. Eldert Van Henten, Waganengin University

High Trellis Twining

• Requires special string and knot for

PNW windy environment

• Tie on “infinitive” long cable, none

similar mechanism usable

• Very large number of knots (>4,000/ac)

done in a short time window

• Operating at a high elevation on

unprepared ground surface with wind

• Courtesy. Qin Zhang. Washington State University

Autonomous Land Vehicles for Demining & Agriculture ALVeDA & MDRD (2010-2013)

Robot Vision: Terrain Classification, RGB-D & Monocular SLAM, Visual Servoing, Soil Estimation in a Bucket Excavator. Field Experiments: Channel mapping in Lahore (left). Scanning a minefield in Beirut (right).

Collaboration: RRLab, TU Kaiserslautern Funding: DAAD, LUMS, National Instruments

Objective: Push performance limits with low-cost vision sensors and simple mechatronics.

Aerial Mapping of Irrigation Canals for Silt Deposition

Guassian Processes (GP) based vol. estimation Proposed System Architecture. Localization and navigation (online), Mapping (offline) Examples of Siltation and bank deterioration in the Indus Basin.

Collaboration: RRLab, TU Kaiserslautern Funding: DAAD, LUMS

Motivation: Automation of annual canal cleaning operation in the world’s largest irrigation network.

Ag Robotics Community

• Emerging area: http://www.fieldrobot.com/ieeeras/
• Summer schools (Sydney 2015), workshops, special issues
• IEEE AgRA (Robotics & Automation Society) technical

committee.

Some Basic Questions …

• Why Automation in developing countries like Pakistan?

– Devolution of governance – Ensuring rights – Conflict resolution

• Major challenges

– Natural resources – Food and Agriculture – Critical infra-structures – Security – Healthcare

Participation Accountability Entitlements

Scaling Problems in Development

Technical and institutional challenges in developing countries are really problems of scales:

• Spatial scales: The inability to monitor and maintain

geographically extent infrastructures (e.g. the world’s largest contiguous irrigation network running over tens of thousands of km of open channels)

• Time scales: The inability to collect information, reconfigure,

and react within short time spans (e.g. irrigation warabandi rosters, issued once in a cropping season despite the fact that water demand and supply varies over much shorter time spans)

• Human scales: The inability to scale human expertise across

institutions (e.g. farmer organization roles in relationship to irrigation officials for maintaining channels, ensuring equity, collecting abiana [water fees], etc.) Robotics, AI, control and automation may be the answer to some

• f these!

An Observation on Scales (Ag Robotics)

• Booming population, urbanization.
• Average farm size in Pakistan going down.
• Can we empower the poor?

1960 2000

How to Engineer the Ag Robotics Revolution for the Developing World?

• Consider the following developments:

– Printing press to personal laser printer – Machine workshop to desktop 3D printer. – Mainframe computing to iPhone.

• Will big-sized PrecisionAg driven corporate farming

give way to small-sized personal Ag Robots?

• A generic, robust and easy

to use platform that can run mechatronic “apps” for common farming tasks.

• Increase productivity of a

small farm to the level of a large mechanized form.