IEEE Robotics & Automation Magazine - December 2014 - 43

Power (W)

Power (W)

equilibrium in case of
abnormal functioning of
the system.
Figure 9 reports the results
acquired during a full walking cycle performed at low
speed and with no carried
weight. The vertical force
components measured at the
feet (which are employed
to trigger the state of the
controller) are reported in
Figure 9(b). The power consumption is shown Fig600 Li
600
S
D
R
S
H
Lo
400
400
ure 9(c), distinguishing 1) the
200
200
total power consumption Pt
0
0
(measured at the power teth4
5
6
0
1
2
3
0
1
2
3
4
5
er), 2) the mechanical power
Time (s)
Time (s)
Pm (that includes mechanical
(a)
(b)
energy only), 3) the power
D: Downward Squat
Li: Lifting Load
dissipation due to Joule losses
R: Rising Phase
H: Handling
Pt Pm Pj
Pe
Pj (estimated via motor curS: Standing
Lo: Lowering Load
rents and motor electrical resistance), and 4) the electrical Figure 10. A plot of the power consumption during a single cycle of (a) squatting and (b)
stand-by power Pe (expended handling a 26-kg cylindrical load. The charts report the total power consumption (Pt), estimated
by the electronic system). Fig- mechanical power (Pm), power losses for Joule dissipation in actuators (Pj), and stand-by power
by the onboard control electronics (Pe) (photos courtesy of the PERCRO Laboratory,
ure 9(d) reports the force adsorbed
TeCIP Institute, Scuola Superiore Sant'Anna).
magnitude measured at the
hands and torso.
Figure 10(a) reports the power consumption data [analo- arm at full horizontal extension. The one-of-a-kind system
gous to Figure 9(b)] relative to squatting with no weight. demonstrated the feasibility of a complex, electrically powFigure 10(b) reports the power consumption data relative to the ered FB exoskeleton with such a target payload during comup-and-down lifting of a 26-kg mass.
mon operating conditions in relevant laboratory settings. BeAs shown, the power consumption of the control electron- cause of its modular hardware and open software
ics is almost task independent and is roughly equal to 120 W. architecture, the obtained system can easily be reproThe maximum registered peak power consumption is approx- grammed and reconfigimately 750 W and occurs during walking (specifically during ured to conduct studies
foot rising). Irrespective of the task, a significant portion of the on complex human-exoThe CCU sets the
power consumption is due to electrical losses in the motor skeleton interaction.
windings (however, these could be reduced significantly by reBesides the successful
instantaneous value of
sorting to high-efficiency brushless torque-motors in place of demonstration, the folthe dc motors adopted here).
lowing limitations of the
the desired velocities at
As it can be observed in the plots, the machine is used at a current version of the
rather slow speed. Several seconds are required to accomplish system have been identithe joint level to impose a
what are considered to be basic tasks. This is, of course, a limita- fied during the experition of the current system and is mainly due to: 1) the lack of ac- mental trials:
velocity at the handle.
tive control of balance, which induces the operator to move ● Operator training: the
slowly through quasi-static equilibrium conditions and 2) even
machine requires sevif the maximum speed of the joints is acceptable, the machine
eral repeated trials for acquiring the necessary skills to
could benefit from improved actuators, at least for a subset of leg
correctly drive it, especially due to the difficulties in
joints (ankle, knee, and hip flexion).
maintaining the equilibrium during heavy load handling
and carriage.
Conclusions
● Slow speed of the machine: The machine is quite slow
A prototype of a medium- to heavy-duty FB-EHPA has been
mainly due to the excessive prudence of the user, who
developed as a research platform for the study of shared
needs to focus too much to keep the system in equilibtransport and handling of heavy loads, up to 50 kg with one
rium, and due to hardware limitations given by the
December 2014

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

43
Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2014
