Robotic Surgery Assisted by Haptic Feedback and Edge AI

Abstract

Robotic surgery has revolutionized the landscape of minimally invasive procedures by offering unparalleled precision, dexterity, and visualization. Yet, one of its most significant limitations remains the absence of tactile sensation, forcing surgeons to rely exclusively on visual cues when manipulating delicate tissues and anatomical structures. This sensory gap can contribute to inadvertent damage, reduced procedural efficiency, and increased cognitive workload. Concurrently, the rise of cloud-based AI for surgical assistance brings powerful analytical capabilities but suffers from unpredictable network latency and data privacy concerns. Edge artificial intelligence (AI)—deploying machine learning models on local computing nodes near the point of care—promises to overcome these challenges by enabling low-latency inference and preserving sensitive data within the surgical suite. Integrating force-reflective haptic feedback with edge AI thus offers a compelling dual-modality approach: restoring the surgeon’s sense of touch while providing real-time, intelligent data processing. In this study, we present a comprehensive evaluation of a prototype robotic surgical platform enhanced with bilateral haptic interfaces and edge-deployed AI algorithms. Our system employs high-resolution force sensors at the robotic end effectors and a lightweight convolutional neural network (CNN) optimized for sub-10 ms inference on an NVIDIA Jetson Xavier NX. A randomized crossover trial was conducted with thirty board-certified laparoscopic surgeons performing three standardized tasks—incision accuracy, knot tying, and object transfer—on anatomically realistic tissue phantoms. Performance metrics included task completion time, force fidelity (measured as the correlation between commanded and rendered forces), positional accuracy, and system latency, alongside subjective workload assessed via the NASA-TLX instrument.