The work aims to combine the subjective and objective weights with the sorting method approaching the ideal solution, and apply them to the product evaluation and optimization system to carry out scientific and effective research sort, so as to solve problem that the subjective weight information accounts for too much in the weight calculation in the current research and optimization of intelligent elderly care auxiliary products. Firstly, Delphi method, questionnaire and other methods were adopted to sort out the needs of the elderly for intelligent assistance products, and propose design evaluation indicators. Secondly, the Analytic Hierarchy Process (AHP) was used to calculate the subjective weight, and the objective weighting method (CRITIC) was used to calculate the objective weights. Then, the formula was used to combine the subjective and objective weights to calculate the comprehensive weight of each index. Finally, the ranking method (TOPSIS) approaching the ideal solution was combined to optimize the design scheme. With three design schemes of intelligent elderly walkers as examples, the framework and process were explained in detail. Practice proved that functional requirements were the most important in the criterion layer of the evaluation system of intelligent elderly care auxiliary products. And in the element layer, easy operation, group interaction and diverse auxiliary functions were important indicators for evaluating intelligent auxiliary products. According to the indicators, the relative closeness of the three schemes was calculated, and the second scheme with the highest closeness was the best scheme. The analyzed product evaluation system, the proposed design strategy and scheme comparison method can provide a reference for the design of intelligent elderly care auxiliary products. Facts have proved that the selection model of the elderly walker design method that comprehensively uses AHP, CRITIC, and TOPSIS is more scientific, and the comprehensive weight value obtained by combining subjective and objective weight methods under multi-dimensional data is more reasonable, which can prove the pertinence and effectiveness of the design.