Modular integrated construction (MiC) represents the most advanced off-site technology. It is challenging to install hefty modules’ safely and effectively in high-rise building projects. Nevertheless, existing crane-lift activities are largely built on the personal experience and subjective judgements of crane operators and signalmen, which often causes time delay and safety hazards. Automatic crane-lift path planning has been demonstrated effective in addressing those issues. However, previous studies seldom considered MiC-specific characteristics such as self-rotation of lifted modules. This paper, therefore, aims to develop an innovative tower crane path planning system for assisting crane operators in high-rise MiC. This system consists of two critical components, i.e., modeling and computing. The modeling component is designed to build three types of models, i.e., original building information models, bounding box models and mathematic models, for setting up the path planning environment. The computing component is designed to work out the optimized crane-lift path using an improved particle swarm optimization algorithm based on three mechanisms, i.e., tower crane operation strategy, fitness function and collision detection. Two real-life MiC projects are used to validate the system. The results indicate that the developed system is effective and efficient in obtaining a collision-free and smooth crane-lift path using limited evolutionary generation and population size. Practically, this study with the advanced crane-lift assistance system should reduce hoisting cycle time and improve safety performance in off-site construction. Scientifically, the paper establishes a theoretical foundation for automated crane-lift path planning and contributes to the application of metaheuristics in the construction industry.