To overcome drawbacks of grey wolf optimization algorithm (GWO), such as low convergence accuracy and easily trapping in local optimum, this paper proposes a hybrid grey wolf optimization (HGWO) algorithm and applies it to the robot path planning (RPP) problem. Firstly, HGWO uses the opposition-based learning method to generate initial population with high qualities. Secondly, the algorithm incorporates its historical information into the individual update method, so as to guide the population evolution. Meanwhile, an elite opposition strategy is applied to explore the space of elite solutions in current population, in order to strengthen the algorithm's exploitation ability. In addition, HGWO adopts the spline interpolation technique to guarantee convergence accuracy and to reduce the optimization difficulty of RPP. Finally, a comparative experiment of function optimization and path planning is carried out. The experimental results show that HGWO has good solution accuracy and robustness.