Dynamic sensor adaptation based on efferent feedback for adaptive bio-inspired sound source localization

Area of Science:

• Neuroscience
• Artificial Intelligence
• Bio-inspired Engineering

Background:

• Auditory perception and localization are crucial for survival, yet artificial systems lag behind biological capabilities, especially in complex environments.
• Efferent feedback, a mechanism in biological auditory systems, is explored as a method to enhance artificial auditory processing.
• Existing artificial auditory systems struggle with dynamic and noisy conditions, limiting their sound localization accuracy.

Purpose of the Study:

• To investigate if sensor adaptation, driven by efferent feedback, can improve sound localization performance in artificial auditory systems.
• To develop and analyze a neuromorphic system architecture inspired by human interaural level difference (ILD) processing and efferent feedback.
• To evaluate the impact of different feedback configurations on the neural processing stage.

Main Methods:

• A neuromorphic system with two bio-inspired acoustic sensors and a neural processing stage (MNTB and LSO neurons) was designed.
• Analysis involved measuring LSO neuron responses with and without efferent feedback across varying ILDs.
• Simulations explored four feedback configurations from LSO neurons to acoustic sensors using synthetically generated sounds.

Main Results:

• Feedback tuning effectively compensated for fabrication-induced mismatches between MEMS sensors.
• Different feedback configurations and simulation parameters significantly influenced the LSO neuron's response to varying ILDs.
• The study quantified the impact of efferent feedback on auditory sensor performance and neural signal processing.

Conclusions:

• Efferent feedback is a viable mechanism for enhancing the performance and robustness of artificial auditory systems.
• The proposed neuromorphic architecture demonstrates the potential of bio-inspired sensor adaptation for improved sound localization.
• Further research into feedback mechanisms can lead to more sophisticated artificial auditory perception systems.