Drones are widely used in remote sensing, agriculture, logistics, and military fields due to their high mobility, low cost, and vertical take-off/landing advantages. Path planning, a core technology for autonomous operations, is limited by traditional GPS/inertial navigation in low-altitude, near-ground, or GPS-denied scenarios. Computer vision can recognize the real-time environment through on-board sensors without prerebuilt maps, so computer vision is a key solution. The integration of computer vision with drone path planning is reviewed in this paper and includes several key components: obstacle detection, visual SLAM, and multi-UAV dynamic path planning. Results show that computer vision has evolved from becoming a supportive sensor to the primary source of information and has drastically improved the safety and performance of drones in unknown and dynamic environments. Current challenges include on-board resource constraints, poor robustness in complex environments, and multi-UAV collaboration difficulties; future efforts will focus on lightweight models, algorithm generalization, and standardized collaboration. This research is crucial for promoting large-scale drone operations in smart cities and low-altitude economies, laying a technical foundation for wider applications in emergency rescue and precision agriculture.
Ziyi Wang (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: