The Economist - USA (2020-02-08)

70 Science & technology The EconomistFebruary 8th 2020

2

1

Leckford Estate, which is run by a large su-

permarket chain. Dick will begin trials later

this year. Tom has already learnt how to

spot broadleaved weeds and is being

trained in the more difficult art of identify-

ing grass-type weeds, which look similar to

cereals. Unlike many vegetables, cereals

are grown more densely in rows, which

makes the mechanical hoeing of weeds a

tricky operation. Armed with Tom’s algo-

rithms, Dick should be able to zap grassy

weeds in cereals accurately.

The company will offer its robots as a

service. Tom will live in a kennel on the

farm, where it will download data for the

farmer and recharge. Dick and Harry will be

delivered to farms as and when they are

needed, much as farmers already bring in

contractors. This business model, reckons

Mr Scott-Robinson, will demonstrate to

farmers that the cost of using agribots will

be competitive with other weed-control

measures and provide additional benefits,

such as being chemical-free.

Dick’s zapping system is supplied by

RootWave, another small British company,

based near Leamington Spa. RootWave,

which calls the process “electricide”, al-

ready makes a hand-held version for spot

weeding. Electricide does not disturb the

soil or harm microbes, says Andrew Di-

prose, the firm’s boss. As for safety, agribots

would—like all farm machinery—need to

be handled by trained operators. Built-in

safety features, such as sensors that detect

an approaching person or animal and shut

the robot down, would reduce risks.

RootWave is working with other compa-

nies interested in electricide. These in-

clude Steketee, a Dutch producer of agri-

cultural equipment. Its tractor-towed hoes

already benefit from computer vision. Ste-

ketee equips one machine with cameras

that detect well-spaced crops, such as vege-

tables, and then moves the hoeing knives

in and out to weed not just along the rows,

but also between individual plants.

Agribots can also hoe. Sebastien Boyer,

an aiexpert who has worked for ibmand

Facebook, co-founded FarmWise, a San

Francisco-based company, in 2016 to devel-

op farming robots. After field trials, the

firm is about to launch its first agribot,

called Titan. This can navigate autono-

mously around a field of vegetables, such

as lettuce, broccoli and cauliflower, identi-

fying individual plants and their location.

As it moves, Titan deploys a series of rotary

blades, which are automatically adjusted

to slice into the soil at an appropriate depth

to dispatch any weeds. Future versions will

do other jobs, such as seeding and micro-

dosing fertiliser and pest treatments di-

rectly to individual plants. “It is personal-

ised care for plants,” says Mr Boyer.

Self-contained agribots will have to

compete with systems towed by smart trac-

tors. Most modern tractors and combine

harvesters can steer themselves across

fields using satellite positioning and other

sensors. Some tractors use digital maps of

crops obtained by satellites and drones to

highlight the places that require fertiliser

or pesticides. Big tractor-producers, such

as John Deere and cnh Industrial, which

makes Case and New Holland machines

(and whose largest shareholder owns

shares in The Economist Group), are devel-

oping fully autonomous tractors.

When chemicals are required on crops,

both tractor-towed systems and agribots

could apply microdoses to the individual

plants that require them, rather than spray-

ing an entire field. Some trials have sug-

gested microdosing could reduce the

amount of herbicide being sprayed on a

crop by 90% or more. basf, a German

chemical giant, is working with Bosch, a

German engineering firm, on a spraying

system that identifies plants and then ap-

plies herbicides in just such a targeted way.

The development of harvesting agribots

is also being driven by ai. These machines

tend to come in many different shapes and

sizes and use a variety of systems specifi-

cally tailored to pick individual crops, such

as tomatoes, asparagus and lettuce. This

suggests that agribots that weed and care

for plants will evolve in a similar manner,

their form depending on the crop con-

cerned and its location, which could be a

field, a vineyard or an orchard. Some may

be smart machines towed by fully autono-

mous tractors, while in other cases swarms

of small agribots will be employed. The one

sure thing is that farmers won’t spend a

penny on any of them unless they prove

they are up to the job. 7

P

erhaps morethan any other, cancer is

seen as a disease of genes gone wrong.

So, as genetic-sequencing technology has

become cheaper and faster, cancer scien-

tists are using it to check which changes to

genes cause tumours to spread.

The latest insights from one group, the

international Pan-Cancer Analysis of

Whole Genomes (pcawg), are revealed this

week in Nature. In an analysis of the full ge-

nomes of 2,658 samples of 38 types of tu-

mour taken from the bladder to the brain,

the researchers give a blow-by-blow ac-

count of how a series of genetic mutations

can turn normal cells into runaway clones.

It provides the most comprehensive analy-

sis yet of where to find this damaging dis-

ruption to dna and, by unpicking the ge-

netics of what makes cancer tick, just how

hard it will be to tame.

For each of the cancer samples, the team

produced a read-out of the tumour ge-

nome—the 3bn or so individual dnalet-

ters—and compared it with the genome se-

quences of healthy cells taken from the

same patients. In this way they could look

for the genetic signatures of the cancer

cells, where specific mutations had warped

the genetic information.

Most mutations in the genome are

harmless. But driver mutations, where ge-

netic changes cause a cell to multiply more

easily and faster than other cells, can trig-

ger tumour growth. Many driver mutations

have been found over the past decade and a

handful have been translated into new

medicines. In a fifth of breast cancers (pic-

tured), for example, a driver mutation in

the gene her2 makes cells produce more of

a protein on their surface that encourages

them to grow and divide out of control. A

series of drugs, including Herceptin, target

this protein, and lead to significantly im-

proved survival rates. The same her2 mu-

tation also appears in some lung cancers,

raising hopes that similar therapies could

work against that disease.

The problem is that most cancers have

multiple driver mutations. Indeed, the

pcawgwork found that on average each

The most comprehensive genetic map of cancers ever made shows how hard they

will be to crack

Oncology

The topography of tumours

It’s complicated