In this paper, we developed an artificial intelligence-based data filtering algorithm to improve the accuracy and reliability of MPU6050 sensor measurements. Using the Edge Impulse platform, we trained and optimized the machine learning model for real-time processing of accelerometer and gyroscope readings. The solution was implemented on the ESP32 with a sampling rate of 600 Hz. The experimental results validated the effectiveness of the solution, highlighting its relevance in applications requiring real-time monitoring
