Exploiting the PIR Sensor Analog Behavior as Thermoreceptor: Movement Direction Classification Based on Spiking...

Area of Science:

• Neuroscience and Robotics
• Sensor Technology and Signal Processing

Background:

• Pyroelectric infrared (PIR) sensors are cost-effective, low-power infrared detectors used in motion control and autonomous systems.
• Existing PIR sensor applications often treat them as simple binary detectors, limiting their potential in complex intelligent systems.

Purpose of the Study:

• To develop and validate novel neural processing architectures for PIR sensors.
• To emulate biological neural functions (presynaptic and postsynaptic neurons) using PIR sensor data.
• To apply these architectures for advanced applications like optical flow detection.

Main Methods:

• Implemented an analog processing circuit mimicking the leaky integrate-and-fire model for PIR sensors to generate spiking rates proportional to IR stimuli.
• Developed an embedded postsynaptic neuron architecture using a spiking neural network matrix and digital processing for sensor data integration.
• Utilized a four-PIR sensor array as input for an optical flow detection case study.

Main Results:

• Successfully demonstrated an analog approach to convert PIR sensor signals into spiking rates, emulating presynaptic neuron behavior.
• Validated the postsynaptic neuron architecture's ability to process and retrieve analog information from spike trains, simulating an optic nerve.
• Achieved effective optical flow detection using the proposed neural processing approach with a four-PIR sensor array.

Conclusions:

• The presented neural processing architectures significantly enhance the capabilities of PIR sensors beyond simple motion detection.
• This work validates the feasibility of using PIR sensors in spiking neural networks for complex sensory processing tasks.
• The findings pave the way for more sophisticated autonomous intelligent devices, robotics, and sensor networks.