Haptic enabled surgical robotic arm

Ayelagbe, T.; Osunyemi, P.; Akanbi, G. · IEEE NIGERCON 2026 · 2026

Robot assisted surgery has transformed minimally invasive procedures by delivering enhanced dexterity, tremor filtration, and three dimensional visualization. However, the near complete absence of force feedback in most deployed platforms remains a critical limitation, compelling surgeons to infer tissue interaction forces from visual cues alone, a deficiency associated with elevated tissue trauma, excess grasping forces, and prolonged training curves. This paper presents an integrative review of three coupled technology domains, force sensing hardware, haptic feedback modalities, and AI based force estimation, evaluating strain gauges, piezoelectric transducers, fiber Bragg grating sensors, and model based observers for sensing; kinesthetic, vibrotactile, and pneumatic tactile modalities for feedback; and deep learning, physics informed neural networks, and Gaussian process regression for sensorless force estimation. Drawing on these analyses, a three layer conceptual framework for next generation force aware surgical robots is proposed, with explicit consideration of deployment feasibility in resource limited environments, including sub Saharan African clinical contexts.

Ayelagbe, T. · IEEE NIGERCON 2026 · 2026

This paper presents a MATLAB simulated pipeline for robot assisted bevel tipped needle steering, aimed at applications such as biopsy, brachytherapy, and targeted drug delivery. It combines four established parts into one closed feedback loop: an assumed modes Euler Bernoulli beam on a Winkler elastic foundation, a nonholonomic bicycle kinematic model, duty cycle bevel steering, and a pole placement PID controller with a saturation aware integral rule. None of these four parts is new on its own. The contribution is how they are wired together: a single explicit coupling rule lets the mechanics sub model drive the real time curvature command, via one calibration constant that converts a beam computed tip slope into a curvature. This exposes something a disconnected simulation would not, that the effective plant gain changes with insertion depth, which a fixed gain PID controller cannot correct for, producing a threefold drop in weak regulation accuracy.