AI-Based Obstacle Avoidance, Line Following and Traffic Light Detection Robot Car Using Arduino Nano 33 BLE Sense with TinyML

Abstract

Autonomous vehicles require intelligent perception and decision-making systems to safely navigate dynamic environments. This paper presents the design and implementation of an Artificial Intelligence (AI) based robotic car capable of line following, obstacle avoidance, and traffic light detection using the Arduino Nano 33 BLE Sense microcontroller. The system integrates traditional sensor-based control with Tiny Machine Learning (TinyML) models deployed directly on the microcontroller. Infrared sensors are used for lane detection, an ultrasonic sensor is used for obstacle detection, and the onboard camera module performs traffic light recognition using a lightweight convolutional neural network. The TinyML model is trained with TensorFlow Lite and deployed with TensorFlow Lite for Microcontrollers. Experimental results show that the proposed system achieves reliable navigation and traffic signal recognition while maintaining low power consumption and real-time performance. The proposed architecture demonstrates the feasibility of implementing AI-based perception systems on resource-constrained embedded hardware.