文档介绍：第 23 卷第 7 期 V ol. s to do intelligent ta sks, since th is apt itude enable s it to w ork in a va riab le e nvironm ent. T his p ape r p re sent s a
new con tro l stra tegy fo r robo t m an ipu lato r, w h ich utilizes visual inform a tio n to d irect the m anip ula tor in its wo rk2
ing sp ace, to p ick up an objec t of know n shape but w ith a rbitra ry po sition and directio n. Dur ing the sea rch fo r an
objec t to be picked up , v ision2based cont ro l by clo sed2loop f eedback, refe rred to as visual servo ing, is pe rfo rm ed to
obta in the m o tion con trol of the m anipula to r hand. The system em p loys a gene tic a lgor ithm (GA ) and a pa ttern
m atching technique to exp lore the sea rch space and e xp lo it the be st solution s by th is sea rch technique. S im ula tion
have been pe rfo rm ed, u sing a two2link p lana r m an ipu lato r, in o rde r to find w hich o ne is the be st fo r rea l2tim e visu2
a l servo ing and the re sults show the effect iveness of the control m ethod.
Keywor ds: visua l se rvoing, pa tte rn m a tching, genet ic a lgor ithm
1 引言( In troduct ion ) 在基于模型的辨识过程中, 工作空间中 目标物