Objective To balance the impact and energy consumption of SCARA high-speed parallel robots during food sorting, trajectory optimization methods are used to improve their comprehensive performance and meet the practical needs of smooth and low-energy-consumption food sorting.Methods Based on the analysis of the entire food sorting system, a trajectory optimization method for food sorting is proposed, combining an improved non-uniform quintic B-spline interpolation with multi-objective SCARA high-speed parallel robots. By introducing virtual path points into the starting and ending paths, a non-uniform quintic B-spline interpolation is used to construct the food sorting trajectory of SCARA high-speed parallel robots. The multi-objective trajectory optimization model is based on the simultaneous minimization of operational impact and energy consumption. The model is solved by a multi-objective particle swarm algorithm integrating external archives, global optimal particles, and inertia weight to achieve the trajectory optimization of SCARA high-speed parallel robots. The operational impact and energy consumption of the proposed method are analyzed through experiments.Results The proposed method effectively achieves comprehensive optimization of operational impact and energy consumption in the food sorting of SCARA high-speed parallel robots, and significantly improves trajectory smoothness and algorithm solving performance. Compared with the pre-optimization method, the operational impact and energy consumption are reduced by more than 50%, and the error at different sorting speeds does not exceed 1 mm.Conclusion The combination of the improved non-uniform quintic B-spline and multi-objective trajectory optimization achieves the comprehensive optimization of operational impact and energy consumption for robots in food sorting.