In order to solve the problems of low working efficiency, high energy consumption loss and strong joint wear during robot operation, a non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ) was proposed to optimize the time, energy consumption and impact trajectory of the dual-arm robot installed on the plate. Firstly, the forward kinematics equation of the robot was established, and its inverse kinematics model was obtained by improving the particle swarm optimization algorithm, and the tight coordination posture relationship of the two arms was deduced. Secondly, the curve of joint space trajectory planning was obtained by deducing the cubic B-spline curve. Finally, NSGA-Ⅱ with elite strategy was used to optimize the time-energy-impact multi-objective of the cubic B-spline curve, and the potential optimal solution in Pareto solution set was screened by fuzzy evaluation method, and the simulation was verified by Adams software. The results show that the time of the solution set selected by fuzzy evaluation method is reduced by 16%, and the energy consumption and impact values are reduced by 11% and 12% respectively, which verifies the correctness and effectiveness of the algorithm. The research results of this paper provide a new research idea for dual-arm trajectory planning.