To cope with the increasingly prominent conflict between the increasing demand for autonomous computing and the limited on-board energy for low-altitude aircraft (LAA), as the key to balance performance and power consumption, dynamic voltage and frequency scaling (DVFS) technology has important research value to improve the endurance and mission capability of LAA. Based on the analysis of LAA heterogeneous computing architecture and dynamic load characteristics, the evolution path of DVFS technology is systematically combed, and the mainstream methods are summarized from traditional heuristic and forward-looking prediction to learning-based adaptive and system-level collaborative optimization for heterogeneous platforms. The advantages and disadvantages of these strategies in terms of real-time guarantee, dynamic response and implementation complexity are compared and analyzed. Taking the aerial-aquatic vehicles (AAVs) as an example, the severe challenges to the existing DVFS technology under extreme conditions are discussed. Considering the shortcomings of current research, the future integrated management framework, which integrates mode perception, lightweight prediction and multi-dimensional resource collaboration, is prospected.