A Machine-Learning-Based Access Point Selection Strategy for Automated Guided Vehicles in Smart Factories

Area of Science:

• Industrial Automation
• Wireless Communication Systems
• Machine Learning Applications

Background:

• Automated Guided Vehicles (AGVs) are increasingly deployed in manufacturing for enhanced mobility and flexibility.
• Wireless communication is crucial for AGV operation, necessitating seamless transitions between multiple Access Points (APs).
• Existing Received Signal Strength Indicator (RSSI) based AP selection methods are unreliable in dynamic manufacturing environments due to signal instability.

Purpose of the Study:

• To develop an advanced AP selection technique for AGVs operating in manufacturing settings.
• To minimize communication downtime during AP switching for reliable AGV monitoring and control.
• To overcome the limitations of traditional RSSI-based methods in unstable wireless environments.

Main Methods:

• Harnessing AGV movement patterns (location, trajectory, orientation) to predict optimal AP connections.
• Utilizing machine learning to learn location-, trajectory-, and orientation-specific RSSI from APs.
• Validating the approach with real-world factory data from a unique dataset.

Main Results:

• The proposed method extends potential communication duration per route by 1.34 times compared to conventional signal strength-based switching.
• Demonstrated significant improvement over standard Wi-Fi drivers in communication stability.
• Reduced the need for extensive manual investigation of wireless propagation paths.

Conclusions:

• The novel machine learning-based AP selection technique effectively enhances AGV wireless communication in manufacturing.
• The method offers a more robust and efficient solution for AP switching than current industry practices.
• Automated evaluation and tuning of the wireless environment streamline AGV adaptation to existing AP infrastructure.