Movement Direction and Distance Classification Using a Single PIR Sensor

Pyroelectric infrared sensors (PIR) are excellent devices to detect human/animal presence with a small form factor and rugged design, for cost-effective surveillance. They are widely used to trigger an intruder alarm and activate household appliances upon the presence of a human. However, the analog output from the sensor is proportional to several spatial and temporal relationships between an object in the field of view of the sensor, the sensitivity of the sensor, PIR lens features, and the environmental heat conditions. This article shows the use of only one PIR sensor in conjunction with signal processing and machine learning algorithms to clearly estimate the presence, direction, and distance of the human movements in a hallway. Simulation results show that more than 99% and 93% accuracy, respectively, in classifying direction and distance of the movements in the field of view of the PIR sensor module. Moreover, hardware and software resources are used to address the limited computational capabilities of commercially available general purpose microcontroller and sensory components to reduce the cost of the entire PIR system module. This article uses only one PIR sensor module with many well-known machine learning and computer vision methods for analog pattern recognition and movement classification. Primarily, this information is used to monitor occupational analysis of a particular floor in a building. However, the application domain can be extended to any one of the earlier mentioned areas in this abstract.