IEEE Robotics & Automation Magazine - December 2015 - 87

Normalized stance time: Stance time refers to the period
when the foot is on the ground, and the stance time of both
feet is expected to be equal. As the walking speeds of different trials vary, the stance time of the prosthetic foot is normalized by that of the intact foot of the same trial.
● Normalized peak GRF: Peak GRF refers to the maximal
GRF during stance. It is associated with a large impact on
the residual limb and may lead to unstable or asymmetric
gaits. To evaluate the effect of the proposed control method, the peak GRF measured in the damping control mode
(robotic) is normalized by that measured in the maximal
damping mode (passive).
The power consumption during one gait cycle was also
measured to evaluate the duration of the onboard battery. The
current is sampled by the onboard current sensor with a sampling rate of 200 Hz. The raw current sensor was low-pass filtered by a third-order Butterworth filter with a cutoff
frequency of 20 Hz.
●

Experimental Results
By wearing the PKU-RoboTPro, the amputee subject can perform natural walking gaits on different terrains in outdoor and
indoor environments (please see the video file in IEEE Xplore).
The terrain transition command was sent by the operator
through a wireless transmitter one step before the transition
takes place. The stride time of the amputees on different terrains with and without damping control are shown in Table 3.
With damping control, the locomotion speeds of the amputees
were increased on all terrains except for SD, which was because the prosthesis had a much larger angle range and the

3.4%

1.1%

11.8%

4.1%

Power Consumption
The power consumption curve during one gait cycle averaged over 30 steps for the three amputees during LG walking, as shown in Figure 6. The power consumption during
the idle state is around
1.4 W. As the proposed
The average power
damping control method
uses human kinetic enerconsumption of the
gy during walking to
produce braking torque,
prosthesis during one gait
the prosthesis does not
consume any additional
cycle is around 3.5 W.
electric power except for
idle power during stance.
The main power consumption period is the swing phase when the prosthesis resets to the equilibrium position under position control, and
the peak power is around 23.3 W. The average power consumption of normal walking during one gait cycle is around
3.5 W. With a 0.28-kg rechargeable Li-ion battery (energy
density 195 W * h/kg, see www.lgchem.com), the prosthesis
can be used for more than 12 h or 20,000 steps.
Ankle Angle Trajectories
The ankle angle trajectories on different terrains, averaged
over 30 steps from three amputee subjects, are shown in

10.6%

CoP Shift (cm)

amputees became more cautious during SD. In addition to the
stride time, three kinds of results were provided, the power
consumption, ankle angle trajectories, and gait indicators.

6
4
2
0

1.3%

5.5%

SD
(a)

6.8%

1.7%

2.1%

4.9%

164.3%

10.7%

14.7%

17.9%

15
10
5
0

0.5%

(b)
5.4%

4.3%

1.6%

7.3%

0.8
0.6

0.5

0.4
0.2
0

SD
(c)

RD
Passive

Robotic

(d)

Figure 8. A comparison of different indicators between the passive and robotic prostheses: (a) CoP shift in ML, (b) CoP shift in AP, (c)
normalized stance time, (d) normalized peak GRF. The experiments were carried out on LG, SA, SD, RA, and RD. The numbers above
the bars express the variation quantity of the robotic prosthesis compared with the passive one. The red number means that the
indicator has been improved, while the black number indicates performance deterioration.

DECEMBER 2015

IEEE ROBOTICS & AUTOMATION MAGAZINE

http://www.lgchem.com

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2015