In recent years, topology optimization (TO) has shown great potential in the optimal design of engineering structures. In this study, a parameterized level set (PLS) method capable of solving the compliance minimization problem with both stress and volume constraints is proposed to improve the design efficiency of nodal connections in large-span latticed structures. 3D topology optimization of assembled hub (AH) joints is conducted based on the proposed method. The performances of the optimal designs are compared with those of the traditional joints in terms of volume, stress distribution and compliance value. Different connection types and complex loading conditions are considered to satisfy the requirements of engineering practice. It is indicated that the proposed PLS method exhibits good stability and high computing efficiency during the optimization of AH joints. Compared with the study cases with a single volume constraint, the maximum von Mises stress is effectively controlled by simultaneously introducing the stress and volume constraints. In addition, the variation in the volume constraint has much greater influence on the compliance results than the stress constraint. The compliance continuously increases when the volume constraint decreases from 100% to 40%, while the difference in compliance is less than 1.5% with or without stress constraints. It is demonstrated that the performances of AH joints are significantly improved after the TO design. The reduction rates of mass are in the range of 25.2 % to 35.2 %, and the compliance value is reduced by 0.5 % to 7.6 % for all the study cases.