An Intelligent Self-Balancing Robot with Integrated Object Detection Using MobileNetV2 and PID Control

This paper presents the design and development of a self-balancing two-wheeled robot capable of detecting hazardous objects in a household environment. The stability issue in two-wheeled robots is addressed using a Proportional Integral Derivative (PID) controller to maintain vertical equilibrium, while object detection is achieved via the MobileNetV2 framework, a convolutional neural network (CNN) architecture. A camera, interfaced with a Raspberry Pi, continuously captures video frames, identifying potentially dangerous objects. Upon detection, both the input frame and the detected object are stored for further analysis. The robot’s balance is maintained through the use of an MPU6050 sensor, which provides feedback on the robot’s angular inclination. This feedback is processed by the PID controller to adjust the direction and speed of the DC motors, ensuring the robot remains upright. The design phase included mechanical system planning, ensuring stability and efficiency in the robot’s operation. The final system integrates both object detection and self-balancing, providing a safe and interactive toy for children. This innovative combination of object detection and self-balancing mechanisms demonstrates the practical application of control theory and computer vision techniques in robotic systems. The system offers insights into the effective integration of PID control and object detection using modern neural network architectures.