Design and Construction of a Two-Wheeled Object Tracking Robot

Abstract

This paper presents the design, implementation, and experimental validation of a self-balancing two wheeled robot (SBWR) governed by a three-state Gain Scheduling Adaptive PID controller integrated with Field Oriented Control (FOC) of Brushless DC (BLDC) motors on a dual-core ESP32 embedded platform. The proposed controller switches among Stable (|e| <2○), Dynamic (2○ ≤ |e| < 6○), and Recovery (|e| ≥ 6○) parameter sets whose boundaries are grounded in experimentally measured sensor noise statistics (σ = 0.42○) and the maximum perturbation angle recoverable by fixed-gain control (6○), respectively. A PCA9548A I2C multiplexer resolves the address conflict betweentwo AS5600 magnetic encoders, enabling simultaneous 1 kHz rotor-angle feedback for both motors. A VL53L0X Time-of-Flight sensor provides closed-loop object-following at 20 Hz, and an INMP441 MEMS microphone supports on-device voice command detection. Across 50 experimental trials, the adaptive controller reduces settling time by 50% (2.4–1.2 s), overshoot by 66.7% (15–5%), tilt RMSE by 68.0% (2.5–0.8○), and fall rate by 83.3% (12–2%) over a Ziegler–Nichols fixed-gain baseline. Object-tracking mean error is 1.3 cm and voice recognition accuracy is 94%.