IEEE Robotics & Automation Magazine - March 2021 - 78

y 0
p0
z x

y 0
p

p1

0

y
p0 0
z x

z x

(a)

y 0
z x

Obstacle

Obstacle
p1

(b)

y 0
z x

p1
p0

p1

(c)

p1

p0

y 0
z x
p0

p1

Obstacle

Obstacle

(d)

(e)

(f)

Figure 7. The preliminary tests of the autonomous navigation performed in laboratory settings. In (a)-(c) and (d)-(f), gray dashed
lines indicate the trajectories the robot followed in environments E1 and E2, respectively. In (a) and (d), the robot moved toward
the goal point without facing any obstacle. In (b) and (e), the robot avoided a fixed obstacle, circled in red. In (c) and (f), a person
crossed the robot's path while it was navigating.

number of procedures performed, mean time spent disinfecting the corridor, average traveled distance during the
procedure, and number of personnel required to boot the
disinfection scenario.
Results and Discussion
Experimental results showing the performance of the
autonomous navigation algorithm in the laboratory settings are reported in Figure 7(a)-(c) and Figure 7(d)-(f)
for environments E1 and E2, respectively. The figure
includes the maps built with the laser rangefinder, the
fixed reference frames, and the trajectories executed by the
robot in all the experimental conditions. Three different
conditions were tested: the movement of the robot without
Table 3. The robot's quantitative performance in a
laboratory setting.

78

*

CT (s)

PL (m)

DE (m)

Without obstacle E1

6.41 ± 1.04

2.58 ± 0.13

0.05 ± 0.03

Static obstacle E1

10.81 ± 1.74

3.83 ± 0.47

0.04 ± 0.05

Dynamic obstacle E1

11.2 ± 2.51

2.82 ± 0.77

0.05 ± 0.06

Without obstacle E2

21.32 ± 1.02

7.79 ± 0.19

0.04 ± 0.02

Static obstacle E2

25.56 ± 2

8.35 ± 0.1

0.05 ± 0.02

Dynamic obstacle E2

39.06 ± 14.64

8.68 ± 2.24

0.06 ± 0.03

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

MARCH 2021

obstacles in its path [Figure 7(a) and (d)], the avoidance of
a static obstacle [Figure 7(b) and (e)], and the avoidance of
a dynamic obstacle [Figure 7(c) and (f)]. As expected, the
robot deviated from the unperturbed path once obstacles
were introduced. Quantitative metrics, defined in the " Preliminary Autonomous Navigation Evaluation " section,
appear in Table 3.
Concerning the first environment (E1), the linear path
the robot followed [Figure 3(a)] constitutes the simplest
condition, and it can be used as ground truth to assess
the other two situations. This experimental condition is
in Figure 7(a). The mean CT required to reach the target
position was 6.41 ± 1.04 s, and the traveled path was 2.58 ±
0.13 m. The error at the end of the movement was 0.05 ±
0.03 m. The introduction of a static obstacle led
the robot to deviate from the straight line and
plan an alternative path. This is highlighted
in Figure 7(b). The robot took more time to
SR (%)
reach the target-more than 10 s-and trav100
eled a longer path (3.83 ± 0.47 m). The DE was
93.75
0.04 ! 0.05 m. In the third condition, a human
subject crossed the ward while the robot was
68.75
navigating. The robot stopped until the dynamic
100
obstacle disappeared. This resulted in an
100
increase in the required time to reach the target,
87.5
keeping the PL comparable with the first case.
IEEE Robotics & Automation Magazine - March 2021

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2021
