New STL-Aware Kalman Filter Boosts UAV Safety Monitoring Accuracy

A team of researchers from academia has developed a novel framework for online state estimation and predictive risk assessment of non-cooperative unmanned aerial vehicles (UAVs). The method, detailed in a paper on arXiv (2607.26527), tackles the challenge of asynchronous heterogeneous sensing and uncertain motion modes. It integrates an interacting multiple-model multi-rate Kalman filter with signal temporal logic (STL) to enable simultaneous low-level state tracking and high-level safety reasoning within a common recursive architecture. The key innovation is an STL-aware time-varying mode transition mechanism that updates model probabilities online using robustness measures derived from formal safety specifications.

Experimental results on a real-time UAV monitoring platform demonstrate that the approach improves estimation accuracy and produces earlier, more informative safety warnings compared to existing methods. The framework also generates multi-step state predictions and probabilistic reachable sets for finite-horizon safety evaluation and risk-triggered warning generation. By embedding safety semantics directly into the inferential process, the system can anticipate unsafe behaviors before they become fully observable, making it highly suitable for real-time UAV surveillance and safety monitoring applications.