Abstract
<title>Abstract</title> <p>To address the accurate control of breast intervention robots, this paper proposes to use binocular stereo camera to detect the needle tip position during the robot's motion, and then perform adaptive fuzzy PID control.The main contributions encompass three aspects: First, a real-time stereo vision inspection workflow was developed, integrating Bouguet correction, semi-global block matching, and a circle detection method based on Hough gradient transformation. This workflow achieves sub-millimeter-level consistent extraction of needle tip 3D coordinates in static environments. Second, an adaptive fuzzy PID controller was designed, employing a dual-input, triple-output inference mechanism. This mechanism enables online tuning of control parameters based on position error and its rate of change. Simulation comparisons demonstrate that compared to traditional PID controllers, the proposed controller reduces overshoot by approximately 35% and shortens settling time by about 20%. Third, experimental validation using an integrated robotic system shows that after error compensation, the average targeting error of the primary feed axis is 1.2 mm, with a maximum error of 3.4 mm, meeting clinical accuracy requirements (<2 mm). This study bridges the technical gap between visual perception and adaptive control in breast intervention robots, providing a deployable solution for advancing toward autonomous and precise breast cancer diagnosis and treatment.</p>