Science - 16.08.2019

SCIENCE sciencemag.org 16 AUGUST 2019 • VOL 365 ISSUE 6454 637

GRAPHIC: ADAPTED BY V. ALTOUNIAN/

SCIENCE

FROM A. PONS-GRAU

ECOLOGY

Changing

nutrients,

changing

rivers

Hard exosuits

These devices can

assist with walking

in therapeutic

settings, but

mismatching of

joints and manual

control of gait can

limit speed.

Soft exosuits

The exosuit of Kim

et al. uses an

algorithm to match

the assistance to a

walking or running

gait. The soft

material fexes with

the user’s joints.

Neural interface

activation

Implanted neural

interfaces may

control robotic

assistance or

even activate

neuromuscular

junctions to restore

gait function.

exosuits, aimed at assisting locomotion in

healthy wearers and at healing or assisting

neurological patients.

How can assistance be seamlessly deliv-

ered with wearable robots and exosuits ac-

cording to the wearer-intended behavior?

The work by Kim et al. shows an elegant and

robust gait classification algorithm based on

transitions between potential and kinetic

energy that can detect whether the wearer

is walking or running. However, the detec-

tion of actual user movement or movement

intent can be much more challenging. When

proportional control of wearable robots with

multiple degrees of freedom is attempted—

for example, in hand prosthetics—a more

elaborated strategy to seamlessly command

the robots is needed.

Under these circumstances, robot com-

mands are usually obtained from decoding

neural and bioelectrical signals by using

classification algorithms. A variety of input

signals have been considered, including both

invasive and noninvasive neural interfaces,

and recorded from either the central ner-

vous system (CNS, the brain and spinal cord)

or the peripheral nervous system (PNS, the

muscle or peripheral nerve electrical activity)

( 4 ). Ideally, a neural interface would require

simple connections to human neural struc-

tures, versatile operation with no calibration,

high spatiotemporal resolution, sampling of

a population of neural cells to associate their

activity to relevant motor functions, and

long-term stability.

None of the currently available recording

and signal processing strategies fully meet

these requirements, and new interface tech-

nologies will need to be developed. Given the

relatively lower complexity of the PNS com-

pared with the CNS, and recent developments

in decoding the neural drive to muscles ( 5 ,

6 ), it is realistic to think that we will witness,

in the next several years, the development of

robust human-robot interfaces to command

wearable robotics based on the decoding of

a representative part of the neural code of

movement in humans. The need for wearable

technologies that minimally alter human

biomechanics will result in a transition from

rigid wearable robots to soft exosuits such as

the one reported by Kim et al., and, eventu-

ally, to implantable neuroprostheses that can

influence or assist human movement. The

need for preserving human neuromechanics

while using assistive technology will likely

lead to implantable and networked record-

ing and stimulation neuroprostheses. Such

devices would implement effective interfaces

to decode the wearer’s movement intent and

influence it when necessary to enhance hu-

man performance ( 7 ). j

REFERENCES AND NOTES

1. J. Kim et al., Science 365 , 668 (2019).


2. J. L. Pons, Ed., Wearable Robots: Biomechatronic
   Exoskeletons (Wiley, 2008).


3. J. C. Moreno et al., J. Neuroeng. Rehabil. 10 , 79 (2013).


4. J. R. Wolpaw et al., Clin. Neurophysiol. 113 , 767 (2002).


5. F. Negro, S. Muceli, A. M. Castronovo, A. Holobar, D. Farina,
   J. Neural Eng. 13 , 026027 (2016).


6. A. Holobar et al., J. Neural Eng. 11 , 016008 (2014).


7. Bidirectional Hyper-Connected Neural System, the
   EXTEND project; https://extend-project.eu.
   10.1126/science.aaw

Phosphorus removal

from freshwater systems

has wide-ranging

ecological consequences

By Carles Ibáñez^1 and Josep Peñuelas2,

E

utrophication—the excessive enrich-

ment of a body of water with nutrients

such as nitrogen (N) and phosphorus

(P)—is Earth’s most widespread prob-

lem for water quality ( 1 , 2 ). Growing

evidence suggests a global trend toward

reversing eutrophication. However, in rivers

and estuaries of developed countries and in

lakes of emerging economies, the ongoing

reduction in nutrient inputs—termed reoli-

gotrophication—is much larger for P than

for N ( 3 , 4 ). Although the rapid emergence

of this phenomenon has hindered detailed

monitoring of the ecological effects, a few

studies have documented an abrupt shift

from green to clear waters and consequently

from phytoplankton to macrophytes as domi-

nant primary producers in response to reoli-

gotrophication in rivers and estuaries ( 5 – 7 ).

However, the improvement in water quality

due to P decline does not imply a return to

pristine ecological conditions, because high

N:P ratios trigger undesirable changes in the

ecosystem ( 8 ).

Understanding of the effects of eutrophi-

cation and reoligotrophication mainly comes

from studies of shallow lakes, which can be

in two alternative states: turbid, dominated

by phytoplankton, and characterized by low

diversity and poor water quality; or clear,

dominated by macrophytes, and character-

ized by higher diversity. Anthropogenic eu-

trophication or reoligotrophication causes

abrupt shifts between these states ( 9 ). Many

American and European lakes have recov-

ered from eutrophication following the

control of phosphorus inputs, providing a

(^1) IRTA, Department of Marine and Continental Waters, 43540
Sant Carles de la Ràpita, Catalonia, Spain.^2 CSIC, Global
Ecology Unit CREAF-CSIC-UAB, 08193 Bellaterra, Catalonia,
Spain.^3 CREAF, 08193 Cerdanyola del Vallès, Catalonia, Spain.
Email: [email protected]; [email protected]
Toward more responsive robots
The soft exosuit reported by Kim et al. represents a transition from more restrictive hard
exosuits to future wearables that will respond to nervous system commands.