Page 112 - Mechatronics for Safety, Security and Dependability in a New Era

Ch20-I044963.fm Page 96 Tuesday, August 1, 2006 5:47 PM
Ch20-I044963.fm
96 96 Page 96 Tuesday, August 1, 2006 5:47 PM

The result of the simulation is expressed in Fig. 5, and the values of the slope and intercept of the
least-squares lines are shown in Table 2.

2
1.8
« 1.6 Table 2 Slopes and intercepts of the least-squares lines.
£ 1.4 • •
M 1.2
-iff*"*—•—*" Air-hockey Algorithm I Algorithm II
Air-hocke y
S 0.6 -•-Algo thm I slope a 0.0399 0.0318 0.0459
" 0.4
0.2 S? -*-Algo thm I I - intercept b 0.1219 0.3757 0.2875
0
10 20 30
time (min.)
Fig. 5 Result of different algorithm.
From Table 2, though intercept of the algorithm I is larger than the others, its slope takes almost the
same value to the air-hockey algorithm. This shows that algorithm 1 gained high sweeping rate at the
initial stage, however, the capability of sweeping task will not change to air-hockey motion algorithm
after. On the other hand, slope of the algorithm II is larger than the others. This result indicates that the
capability of sweeping task of algorithm II is much better than the others.
To search this background, sweeping motion was analyzed in detail.

(a) Completed sweeping area. (b) Trajectory of (a).
Fig. 6 Completed sweeping area and its trajectory.
Fig. 6 (a) shows the area of completed sweeping task of air-hockey motion of 40 min. The place of
gradation in circle is the position of the robot ended up with. Fig. 6 (b) shows the trajectory of (a). With
air-hockey motion, places around the border of the rooms tend to left un-swept from Fig. 6 (a).
Sweeping along the wall at the beginning solves this problem. In addition, it seems that there is an
effect to raise the sweeping rate by avoiding the overlaps.
On the other hand, Fig. 6 (b) testifies to presence of unevenness in the task. The robot cannot slip out
from the places in the case when the robot enter the place where the entrance is small, and that causes
the robot move around the same area. It is considerable that probability of the robot slipping out from
the space written above would rise if it switches to wall following motion regularly. And as a result, the
efficiency of sweeping task would rise.

VII. CONCLUSION
In this paper, we analyzed and evaluated the sweeping algorithm of reactive method on the purpose of
developing efficient sweeping algorithm. A simulator was made to compare and to evaluate the
algorithms in the same condition. The effects of the algorithm of commercial robot were estimated by
comparing the algorithm of basic air-hockey motion and the algorithm of the commercial robot.

REFERENCES
[1] J. Ota, D. Kurabayashi, T. Arai (2001). Introduction to Intelligent Robots, Corona, JP.
[2] S.P. Engelson et al. (1992), Error correction in mobile robot map learning, 1CRA, pp. 2555-2560.
[5] T. Lozano-Perez (1983), Spatial planning: a configuration space approach, IEEE Trans. Comp. 32, pp.
108-120.
[3] http://www.botlanta.org/
[4] http://www.servicerobots.org/cleaningrobotscontest/index.php