Multi-objective optimization problems (MOPs) are significant in real world and often solved by using multi-objective evolutionary algorithms (MOEAs). However, the existing MOEAs are all facing challenges of falling into local optimization, low convergence speed and uneven distribution. To solve the above challenges, this study proposed a novel algorithm called altruistic NSGA-II (ANSGA-II), which embeds the central idea of altruism into NSGA-II. In the procedure, nurturing cost is self-adaptively composed by Pareto cost and crowd cost to better contribute to different periods in iterations. Besides, the abandonment threshold is also self-adaptive according to the abandonment situation of last generation, which accelerates convergence speed and assists population in escaping from local optimization. Moreover, double selections strategy consisting of k-nearest neighbor selection and non-dominated selection helps to balance convergence and diversity of population. The experimental results determine optimal ranges of parameters and validate the utility of each strategy. The comparisons with other algorithms demonstrate the great competitiveness of the proposed algorithm.