Disrupting Agriculture With Flying Robots

Daryl Wilding-McBride, CTO @ DiUS

Building new capabilities at DiUS

CASA Certified UAV Controller

Building autonomous flying robots since 2013

Co-founder of Flying Robot School

About me

Needs of agriculture

Technology of flying robots

Applying the technology to the needs

The reality

Thoughts on disruption

About this talk

The needs of agriculture

60 Mpeople are fed by Australian farms

250 Mpeople in China’s middle class want

Australia’s safe and clean food

<1%of Australia’s land is irrigated agriculture

Farmers need to be

2%more efficient each year for status quo

Source: ABS 2011 Census of Population and Housing

Ageing farm workforce

Source: Adoption of precision agriculture-related practices, Rick Llewellyn & Jackie Ouzman, CSIRO, 2014

The technology offlying robots

Terminology

Remotely Piloted Aircraft (RPA)

Unmanned Aerial Vehicle (UAV)

Drone

Flying Robot

3DR Solo DJI Phantom 3

Autonomous missions Autonomous missions (3rd party GCS)

On-board and GCS programmable GCS programmable only

Easy to add sensors Hard to add sensors

Open source Closed source

Can detach camera Can’t detach camera

Decreased cost

Decreased power requirements

Increased capability

Compute

Sensors

Technology changes making it possible

AccelerometerGyroscopeCompassBarometerI2C interface$40

https://www.adafruit.com/product/1604

GPS receiver10 Hz66 channels< 3 m accuracySerial interface$53

https://www.adafruit.com/product/746

All this means...

Flying robots in agriculture

Multispectral crop survey

Farm scouting

Counting stock

Scaring birds

Roles for RPAs on the farm

Benefits of RPAs

Wide area coverage

View from above

More frequent

More detail: 3cm GSD cf. 30m

Not limited by cloud

What is NDVI?

Source: Agribotix

The post-trial phase

Disruption?

Early adopters are experimenting

It’s fun, promising

Disruption?

Early adopters are experimenting

It’s fun, promising

However...

Not entrenched

No evidence of changing behaviour

CASA rule update

● one RPA at a time● visual line of sight● < 120m AGL● daytime only● > 30m from people● > 5.5km from a controlled aerodrome; and● not over a public safety operation

CASA rule update

● one RPA at a time● visual line of sight● < 120m AGL● daytime only● > 30m from people● > 5.5km from a controlled aerodrome; and● not over a public safety operation

BVLOS Unsupervised

Human operator monitors flight No human operator

$2,880 and 28 days for assessment

RPA to have

● ADS-B● Navigation lights● Strobe lights● Battery backup● Dual communications

BVLOS requirements

BVLOS requirements (continued)

● Area approval● CASA to observe simulation session● Traffic density

● Redundant everything● Who’s watching?● Who will intervene?

Unsupervised flight

Unsupervised flight paused for now

Watching brief on BVLOS regulation

Imagery understanding with Deep Learning

What’s next

On disruption

Things we can control

● Usability● Making information actionable

What needs to happen for widespread disruption

What needs to happen for widespread disruption

Things we can’t control

● Approval of unsupervised flight● Farm-wide fast internet

Will satellite flocks disrupt drones?

Planet Labs - 5m GSD, every day

For crop survey, probably

For scale

Much less fiddly

It’s not just competition; it’s creating new models

On disruption generally

It’s not just competition; it’s creating new models

Usability and actionability are critical

On disruption generally

It’s not just competition; it’s creating new models

Usability and actionability are critical

Regulation resists disruption

On disruption generally

It’s not just competition; it’s creating new models

Usability and actionability are critical

Regulation resists disruption

Historical innovation

On disruption generally

Staying in touch

Daryl Wilding-McBride

@DarylWM

dwm@dius.com.au