Published Paper Details:

Line-following robots remain a canonical bench- mark for closed-loop control on embedded platforms because they combine non-linear sensing, actuator saturation, and tight real-time constraints. Although proportional-integral-derivative (PID) control is widely used, deployments frequently suffer from oscillation on sharp curves, integral windup under actuation limits, and sensitivity to surface reflectance changes. This paper presents a curvature-aware discrete PID design for differential- drive line followers that (1) uses a filtered centroid-based error estimator from a reflectance array, (2) incorporates derivative filtering and back-calculation anti-windup for stability under PWM saturation, and (3) applies lightweight gain scheduling driven by an online curvature proxy derived from the sensor profile. We provide a reproducible modeling pipeline, embedded implementation details for 1 kHz control, and an evaluation pro- tocol across three track types (low-curvature, mixed, and high- curvature). Results show that the proposed controller reduces mean absolute lateral error by 3.6% versus a fixed-gain PID and by 44.6% versus a Ziegler-Nichols tuned PID, while improving the 95th-percentile error and reducing oscillatory behavior at high speed. All figures are generated from included scripts, and all references are provided in a BibTEX database