Page 111 - Mechatronics for Safety, Security and Dependability in a New Era

Ch20-I044963.fm Page 95 Tuesday, August 1, 2006 5:47 PM
Tuesday, August
5:47 PM
1, 2006
Ch20-I044963.fm
Page 95
95 95
IV. ANALYSIS OF AIR-HOCKEY MOTION
First, the basic air-hockey motion algorithm was installed in the simulator, and the transitions of
sweeping rate for each room are observed. Data were taken 100 times for each room and averages of
those were calculated. Results of those rooms are shown in Fig. 3. The plots indicate the percentage of
the sweeping rate at certain time in the same room, starting at the same place.
According to the plots, 90% of sweeping was generally achieved in around 20 min., and 95% was done
in about 40 min. in 12 mat rooms. As the transition curve can be approximated into logarithm curve,
the performance of the task can be quantitatively estimated using logarithm of the rate of un-swept area
5;
S = - log(l - sweeping rate) ~at + b (1)
The slope a signifies the quickness of the sweeping task, and the intercept b shows the initial boost of
the sweeping rate. Each value obtained from simulation is shown as Table 1.

Table 1 Values of slopes and intercepts.
Rectangle L-shape T-shape
slope 0.033 0.029 0.032
intercept 0.211 0.189 0.187

U-shape Step shape Constricted
slope 0.033 0.033 0.032
intercept 0.158 0.222 0.210
Fig. 3 Sweeping each room in air-hockey motion

V. EFFECT OF CHANGING INITIAL POSITION
The change in the initial position of the robot may affect on the sweeping rate. Here, by starting from
different initial position shown in Fig. 4 (a), influence to the changes of sweeping rate is researched.
To make the condition as different as possible, initial position was set at the center of the room distant
from the obstacles, under the table where the obstacles concentrate, and at the corner of the room near
the obstacle. Sweeping task was executed 100 times using air-hockey motion from each initial position
to see the changes of sweeping rate. Fig. 4 (b) shows the sweeping rate average of each initial position.
From Fig. 4 (b), amplitude is settled in 9% at each moment, and considerable change in sweeping rate
by changing the initial position of the robot couldn't be seen from the result.

1.6
1.4
11-2
» 1
•|0.8
Comer
£0.6
§0.4 —~/^- -• - Under the table
0.2 r -*-Center
0
0 10 20
time (mirO
(a) Initial positions used in the simulation, (b) Sweeping point starting from different initial position.
Fig. 4 Initial positions used in the simulation.
VI. ANALYSIS OF COMERCIAL ROBOT ALGORITHM
In this section, effects of the motions when combining motions installed on commercial robot cleaners
with air-hockey motion are analyzed. Therefore the motions that were mounted on the robot cleaners
were imitated on the simulator. The motion algorithms mounted on the simulator are as follows.
Algorithm I; robot goes straight to the wall, and make a round trip along the walls of the room.
Then switch to air-hockey motion.
Algorithm II; first, take air-hockey motion for 60 sec. Continue the motion till it collides to the wall.
Then switch to the wall following motion for 30 sec. After that, it parts from the wall in random
angle, and switches to the air-hockey motion again. Those two algorithms are repeated regularly.