Intelligent Monitoring and Predictive Modeling for Six-Axis Robotic Arms Using Multi-sensor Fusion, LSTM-Based Trajectory Prediction, and Random Forest Temperature Forecasting

This study presents an intelligent monitoring and predictive maintenance framework for six-axis robotic arms using multi-sensor data fusion, Long Short-Term Memory (LSTM)–based motion trajectory prediction, and Random Forest (RF)–based thermal forecasting. Rather than relying on static threshold alerts, the framework achieves intelligent monitoring by autonomously processing synchronized multi-sensor streams, dynamically learning non-linear dynamics, and proactively forecasting trajectory and thermal degradation trends. A custom sensing module with an MPU-6050 IMU and an Arduino Mega 2560 controller synchronously acquires real-time kinematic and thermal signals from MG996R servo motors to ensure stable condition awareness.

To address equipment failures and performance degradation caused by load variations and heat accumulation, data-driven predictive models are deployed. The LSTM network captures non-linear temporal dependencies in joint motion, achieving coefficients of determination R² up to 0.95 for Pitch and Yaw axes on 2,304 motion records. Concurrently, a hyperparameter-optimized RF model forecasts thermal dynamics, outperforming Linear Regression and k-Nearest Neighbors. Trained on 10,032 samples, the RF model attains an R² of 0.9949 and a mean squared error of 0.004, demonstrating high sensitivity to operational time and load conditions.

The proposed framework enables early-stage anomaly detection and provides a scalable predictive maintenance foundation for smart manufacturing. Platform scalability is realized through a modular I2C multiplexed hardware architecture and standardized data preprocessing pipelines, allowing the infrastructure to seamlessly expand to additional joints, multi-robot workcells, or alternative topologies without structural re-engineering. This framework enhances operational efficiency, system reliability, and fault resilience in Industry 4.0 robotic systems.