This paper presents an intelligent path planner for lifting tasks by tower cranes in highly complex environments, such as old industrial plants that were built many decades ago and sites used as tentative storage spaces. Generally, these environments do not have workable digital models and 3D representations are impractical. The current investigation introduces the use of cutting edge laser scanning technology to convert real environments into virtualized versions of the construction sites or plants in the form of point clouds. The challenge is in dealing with the large point cloud datasets from the multiple scans needed to produce a complete virtualized model. The tower crane is also virtualized for the purpose of path planning. A parallelized genetic algorithm is employed to achieve intelligent path planning for the lifting task performed by tower cranes in complicated environments taking advantage of graphics processing unit technology, which has high computing performance yet low cost. Optimal lifting paths are generated in several seconds.