A brain-inspired robot pain model based on a spiking neural network

Area of Science:

• Robotics
• Neuroscience
• Artificial Intelligence

Background:

• Pain is vital for organism survival and self-preservation.
• Existing robot pain models focus on human pain recognition or obstacle avoidance, lacking adaptive danger response.
• Robots lack inherent pain capacity and adaptive self-preservation mechanisms.

Purpose of the Study:

• To develop a computational model of robot pain inspired by biological pain mechanisms and the Free Energy Principle (FEP).
• To enable robots to quantify internal injury and adaptively respond to danger.
• To explore the fundamental nature of pain through computational simulation.

Main Methods:

• Constructed a Brain-inspired Robot Pain Spiking Neural Network (BRP-SNN).
• Incorporated spike-time-dependent plasticity (STDP) learning rule and population coding.
• Quantified machine injury by detecting multi-modality sensory information coupling.

Main Results:

• The BRP-SNN model demonstrates biological interpretability through comparative analysis with neuroscience experiments.
• Successfully tested on real robots for alerting actual injury and preventing potential injury tasks.
• Generated "robot pain" as an internal state to represent machine injury.

Conclusions:

• The BRP-SNN model offers a novel approach to integrating pain concepts into intelligent robotics.
• Provides a new computational perspective for exploring the nature of pain in cognitive neuroscience.
• Enhances robot self-preservation capabilities and longevity through adaptive responses to injury.