The work aims to propose an improved genetic algorithm for robot path planning, in order to improve the local traps and premature convergence of the traditional genetic algorithm in the path planning of palletizing robot, as well as the energy consumption and path smoothness of the robot. Firstly, aiming at the problems of traditional genetic algorithm, the algorithms and methods of population initialization, fitness function, selection operator, crossover operator and mutation operator were adjusted and improved, and the excellent algorithms were fused. Aiming at the problem that the basic genetic algorithm mainly focused on the shortest path and thus ignored the energy consumption and path smoothness of the robot, a fitness function which took into full account the control of distance and turning times was proposed. Finally, the improved algorithm was applied to the path planning of the palletizing robot. The simulation results showed that, compared with the basic genetic algorithm, the proposed algorithm could find better path quality. Not only the distance was shorter, but also the turning times were much less than other algorithms, and the path was smoother, which proved the effectiveness of the algorithm. The path of palletizing robot based on the proposed algorithm is smoother while taking into account the optimal distance. Because of the reduction of turning times, the energy consumption of the robot is lower. At the same time, the simulation results show that the real-time performance of the algorithm is better.