Abstract: In order to improve effect of fruit fly optimization algorithm (FOA) used in path planning of welding robot and overcome its weakness, improved fruit fly optimization algorithm (IFOA) was proposed in this paper. Compared with FOA, IFOA has two improvements. The first is that using adaptive step to instead fixed step in FOA. The second is that integrating the explosion operation of the fireworks to FOA. Thus better search ability can be obtained by those two improvements. In the practical application of path planning of welding robot, the results obtained by IFOA has short welding path and better stability than FOA.