Abstract:

Targeting the challenges of complex and volatile environments, difficulties in multi-Unmanned Aerial Vehicle(UAV) coordination, and insufficient dynamic risk avoidance in low-altitude emergency rescue, An intelligent collaborative control technology based on multi-dimensional situational awareness is proposed. By constructing a heterogeneous information fusion model based on dynamic bayesian networks, airborne sensor data and external spatio-temporal situational data are mapped into a 3D dynamic risk map. On this basis, a risk-coupled multi-agent reinforcement learning scheduling strategy and an adaptive path planning method are designed, to integrate real-time risk quantification into the decision-making and planning loop. Field experiments demonstrate that this method significantly improves the perception accuracy of hidden dynamic risks, increases mission success rate, shortens obstacle-avoidance response time and overall mission completion time, and reduces risk exposure duration in complex environments, thereby providing technical support for safe and efficient low-altitude emergency rescue systems.

Key words: low-altitude emergency rescue, multidimensional situational awareness, UAV swarm, intelligent collaborative control, dynamic bayesian network