Activity Recognition for IoT Devices Using Fuzzy Spatio-Temporal Features as Environmental Sensor Fusion

Area of Science:

• Computer Science
• Artificial Intelligence
• Ubiquitous Computing

Background:

• The Internet of Things (IoT) is rapidly evolving with new devices and sensors offering enhanced precision in daily life.
• Integrating heterogeneous sensors for activity recognition presents challenges due to data variability and computational demands.

Purpose of the Study:

• To propose a general descriptor for heterogeneous sensors using spatial-temporal features and fuzzy logic.
• To enable low-power devices to perform learning and evaluation tasks for activity recognition.
• To demonstrate the methodology's potential in a real-world intelligent environment.

Main Methods:

• Utilizing fuzzy logic to extract spatial-temporal features from diverse sensor data (UWB, inertial, wearable, binary).
• Developing a fuzzy sensor representation for efficient data fusion.
• Applying light and efficient classifiers for activity recognition.

Main Results:

• The fuzzy logic-based methodology successfully fused data from heterogeneous sensors.
• Encouraging performance was achieved in recognizing inhabitant activities.
• The approach proved efficient for devices with limited computing power.

Conclusions:

• Fuzzy logic provides an efficient method for generating spatial-temporal features from heterogeneous IoT sensors.
• This approach facilitates accurate activity recognition on resource-constrained devices.
• The methodology shows significant potential for practical applications in intelligent environments.