Simulation Fitting Method for Setting Motion Parameters of Robotic ...

Table 1 Error comparison
Linear error
Square error
JLM SFM JLM SFM
1 282.7539 -31.6493 117.5796 74.4139
2 299.8585 -18.6237 135.8531 86.8413
3 244.7837 -10.0649 116.8211 75.6017
4 169.971 11.6239 71.1852 42.3516
5 59.8206 40.3002 33.7386 18.1548
6 45.0723 0.574 39.2045 20.1163
7 8.744 -4.4178 34.7495 15.3578
8 -63.6139 21.0387 22.7306 8.5813
9 -101.151 50.1628 24.9156 11.4755
10 -150.719 56.1411 36.8875 19.145
11 -176.029 53.2019 53.8563 32.5745
12 -241.814 42.7809 85.025 52.4728
13 -282.754 30.299 117.5796 74.5112
14 -299.859 18.684 135.8531 86.8327
15 -244.784 10.0424 116.8211 75.6027
16 -167.971 -11.6179 71.1852 42.3517
17 -59.8206 -40.2987 33.7386 18.1547
18 -45.0723 -0.5738 39.2045 20.1164
19 -8.744 4.4176 34.7495 15.3578
20 63.6139 -21.04 22.7306 8.5813
21 101.1514 -50.1645 24.9156 11.4757
22 150.7191 -56.146 36.8875 19.1454
23 176.029 -53.1993 53.8563 32.5744
24 241.8141 -42.7759 85.025 52.4725
From table 1, we can see, on 24 discrete time points, the errors with simulation fitting method
are all smaller than those with JLM. The linear error of the form er ranges 1.27%~67.37% of the
latter, with an average of 26.26%; the rate with square error i s 7.75% ~64.71% , with an average of
54.89%。
6 Conclusion
This article promotes an innovativ e method, simulation fitting method, in orde r to reduce
the difference betwee n link and fish curve. Two points feature this me thod : first, it is borrowed
from physics phenomenon, so we need n’t to worry about its astringency an d stability; second,
and give out adjusting direction in simulation, and therefore avoid unacceptable calculation
amount. Mo reover, this method can also be used whenever l ink is used to tak e the form of curve.
Nevertheless, we didn’t test its performance with fluid mechanics, but most likely, the result