Researchers have developed a new data-driven framework for adaptive safety filtering in mission-critical intelligent systems. This method learns a safe operating envelope from data and updates it in real-time to account for time-varying limitations. The approach utilizes a radial basis function-kernel support vector machine to initially define the safe set, which is then contracted using a continuous-time decremental SVM update law based on a degradation signal. A homotopy-smoothed SVM-CBF is introduced to manage transitions in the learned barrier, and a quadratic-program-based safety filter enforces the time-varying barrier under degraded input constraints. Simulations on a vertical takeoff and landing model demonstrated the framework's ability to maintain safety and avoid abrupt barrier-switching effects. AI
IMPACT This research could improve the reliability and safety of autonomous systems operating in dynamic environments.
RANK_REASON Academic paper detailing a novel control theory framework. [lever_c_demoted from research: ic=1 ai=1.0]
- Control Barrier Function Based Human Assist Control
- homotopy-smoothed SVM-CBF
- radial basis function
- support vector machine
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
