Answers are grounded in this paper and cite their sources.
Traditional navigation systems for unmanned aerial vehicles (UAVs) rely on GPS signals, which can be susceptible to jamming or spoofing under electronic warfare conditions, in densely populated urban areas, or due to natural electromagnetic interference. This vulnerability can lead to disorientation, mission interruptions, and even the loss of the vehicle. An alternative to GPS is visual navigation, which utilizes cameras and image processing algorithms. However, this method faces difficulties in rapidly changing environments without satellite signals. Purpose: to analyze the evolution of visual navigation methods for UAVs in scenarios lacking GPS signals. This will include a discussion on key technologies, their integration with inertial navigation systems and artificial intelligence, as well as an assessment of their benefits, challenges, and application prospects, with an emphasis on collaborative approaches in multi-agent systems. Results: there is an increasing interest in visual navigation, with a focus on VSLAM for estimating the posterior probability of trajectories and maps, visual odometry aimed at minimizing reprojection errors, and the integration of multi-sensor fusion through extended Kalman filters to achieve meter-level accuracy in complex environments. The incorporation of artificial intelligence, especially convolutional neural networks, enhances resilience to illumination variations and facilitates real-time adaptability. In the context of multi-agent systems, cooperative SLAM models with correspondence matrices among agent maps have demonstrated the capability to reduce the root-mean-square positioning error to below 1 meter in simulations, even amid intermittent communication and loss of GPS. Practical significance: these findings enhance UAV autonomy in scenarios lacking GPS signals, such as searchand-rescue operations in urban areas, monitoring of agricultural lands, environmental control, and specific tasks related to electronic warfare, thereby ensuring coordinated swarm actions for efficient territorial coverage and risk mitigation.
No takes yet. Share an insight, caveat, or question.
Sergey Zlobin (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: