A Dual-Arm Robotic System With Virtual Fixture-Guided Light Tracking for Fundus Surgery

Abstract

Retinal surgery demands submillimeter precision in constrained intraocular spaces, with subretinal injections requiring accurate bimanual coordination between surgical tools and illumination devices. This paper presents the Dual-Arm Robotic System for Subretinal Injection (DASSI), featuring a motion-decoupled Remote Center of Motion (RCM) mechanism and an initial placement strategy to optimize dual-arm workspace overlap. An Automatic Light Pipe Tracking algorithm with Guided Virtual Fixtures (ATLP-GVF) is developed to eliminate redundant light pipe motion when illumination needs are met. Integrated with its hardware design (motion-decoupled RCM mechanism) and ATLP-GVF, DASSI achieves precise performance: orientation tracking error ( $0.65^{\circ } \pm ~0.28^{\circ }$ ), depth error (1.86 mm $\pm ~0.93$ mm), and stable illumination (tracking offset ratio 26% $\pm ~14$ %). Notably, in the ex-vivo experiment, ATLP-GVF reduces light pipe movement duration to less than 10% of that with continuous tracking methods, representing a reduction of more than 90%. DASSI demonstrates potential to address bimanual coordination challenges in retinal microsurgery, offering a precise platform for complex procedures such as subretinal gene-therapy delivery.