New Scientist - USA (2021-12-18)

18/25 December 2021 | New Scientist | 13

PA

BL

O^ N

AV

AR

RO

• LO

RB

ÉS

Some dinosaurs could

run at 45 km per hour

Palaeontology

Carissa Wong

of ostrich-like animals called

the ornithomimids, are thought

to have reached top speeds of

about 48 kilometres per hour.

The newly discovered tracks

offer a rare insight, as most

fossilised dinosaur footprints

are made by walking, rather

than running, dinosaurs.

“High-velocity trackways are

very scarce around the world,”

says Torices. But recent findings

are changing the image of

dinosaurs from just walkers

to very good runners, she adds.

In line with previous

discoveries, this study shows

that the fastest dinosaurs were

medium-sized animals, says

John Hutchinson at the Royal

Veterinary College in the UK.

When the researchers looked

more closely at the estimated

speeds between consecutive

footprints, they found evidence

that the second track was

produced by an agile dinosaur

that could suddenly reduce its

speed before rapidly moving on.

The dinosaurs may have been

pursuing prey such as iguanodon,

a horny-beaked herbivorous

dinosaur, says Torices. ❚

A footprint left by a

dinosaur moving at high

speed, found in Spain

ABOUT 100 million years ago

in what is now northern Spain,

dinosaurs reached running

speeds of up to 45 kilometres

per hour, making them among

the fastest known dinosaurs.

The conclusion comes from an

analysis of fossil footprint tracks.

A few three-toed dinosaur

footprints were discovered at the

site in La Rioja some 35 years ago.

Now, further excavations at the

site have revealed more prints.

In total, there are five footprints

forming one short trail, and a

second trail of seven footprints.

The trails were made by

two medium-sized dinosaurs,

each about 2 metres tall and

4 to 5 metres long, that may have

belonged to either the spinosaur

or carcharodontosaur family

of predatory theropods.

Angélica Torices at the

University of La Rioja and her

colleagues analysed the distance

between consecutive footprints

made by the same leg, or the

stride length. They then plugged

the information into equations

previously shown to describe

the relationship between stride

length and speed across a range

of vertebrates, including humans.

This revealed that the dinosaurs

were among the fastest we know.

The first track was made by a

dinosaur that gradually accelerated

to speeds of 37 kilometres per

hour, while the second reptile

was moving at up to 45 kilometres

per hour with abrupt changes

in speed, suggesting it switched

directions as it ran (Scientific

Reports, doi. org/ g8s6).

The fastest dinosaurs, most

of which belonged to a group

5 metres

Estimated body length of the

fast-running predatory dinosaurs

A TABLE TENNIS-playing robot

can keep up a rally against

humans, but like many amateur

players, it struggles when

attempting fancier shots.

Yapeng Gao, Jonas Tebbe and

Andreas Zell at the University

of Tübingen in Germany

designed a computer simulation

in which a virtual robot arm

equipped with a table tennis

racket attempted to return ping

pong balls across a virtual table

tennis table.

The researchers ran this

simulation so that a machine

learning algorithm could

work out how the velocity and

orientation of the racket affects

the path the ball takes.

Once this algorithm, which

learns by trial and error, could

reliably return the ball, the

researchers set it up to control

the movement of a real robot

arm positioned next to a

table (pictured).

The system used two cameras

to track the location of the real

ball every 7 milliseconds, and the

algorithm processed the signals

and decided where to move the

robotic arm to hit and return

the ball.

The signals that the algorithm

sent allowed the robot arm to

accurately play shots to within

an average of 24.9 centimetres

of the intended location. This

accuracy level was slightly

worse than when the algorithm
was working with a simulation –
a common occurrence, says
Tebbe, as computer simulations
can’t accurately represent
everything in the physical world.
The entire process – including
training in the virtual simulation
and in the real world – took just
1.5 hours, demonstrating how
rapidly algorithms can learn to
perform in a new situation
(arxiv.org/abs/2109.03100).
However, although the robot
performed well against human
players, it was tripped up by fast
shots – and, surprisingly, by
slow ones. “If a ball is slow, the
robot needs to generate more
speed,” says Tebbe. It struggled
to do that, and the ball often
slumped off the racket.
“By training the system for
a relatively short period of time
the robot is able to cope well
with differences in serve, and
capable of returning using a
random policy,” says Jonathan
Aitken at the University of
Sheffield in the UK, who
wasn’t involved in the study.
Aitken is surprised the
algorithm flunked returning
slow shots. He also finds it
interesting that it sometimes
struggled with making shots
because of the mechanical
limitations of the robot
system, rather than because
of shortcomings with the
algorithm.
The robot arm has other
limitations. For instance, it
struggles to play backspin
shots, says Zell, because it is
unable to hold the racket at
the required angle needed to
perform such shots. But despite
these issues, he believes the
robot is a good player.
“It’s not worse than a regular
human player,” he says. “It’s
GA already on par with me.” ❚

O,^
TE
BB
E,^ A

ND

ZE

LL.

20

21

.

This robot arm can play

table tennis as well as

a regular human player

Technology

Chris Stokel-Walker

Robot learns to

play table tennis

in just 1.5 hours