Dual‐Modal Safety Framework for Robotic‐Assisted Bronchoscopy via Endoscopic Vision and Haptic Feedback

Background Teleoperated flexible bronchoscopy lacks distal haptic feedback, increasing the risk of unobservable iatrogenic tissue trauma during endoluminal exploration. Methods This study proposes a robot‐assisted bronchoscopy system integrated with a dual‐modal safety framework. A visual module utilises image moments to provide predictive alerts for collision avoidance. Concurrently, a Transformer‐based network estimates distal interaction torques from proximal actuation, enabling real‐time haptic perception at the leader console. Results Phantom validations across fragile tissue rupture, sidewall approximation, and luminal exploration tasks demonstrated the framework's efficacy. The visual warnings helped decrease excessive force contacts, while the learning‐based estimator delivered reliable distal force perception. Conclusions The proposed dual‐modal teleoperated safety architecture resolves the trade‐off between operational dexterity and safety in flexible bronchoscopy. By integrating predictive visual warnings and real‐time haptic feedback, the system improves the operator's situational awareness, decreasing the risk of iatrogenic trauma during the operation.