Abstract:Bees display the remarkable ability to return home in a straight line after meandering excursions to their environment. Neurobiological imaging studies have revealed that this capability emerges from a path integration mechanism implemented within the insect's brain. In the present work, we emulate this neural network on the neuromorphic mixed-signal processor BrainScaleS-2 to guide bees, virtually embodied on a digital co-processor, back to their home location after randomly exploring their environment. To realize the underlying neural integrators, we introduce single-neuron spike-based short-term memory cells with axo-axonic synapses. All entities, including environment, sensory organs, brain, actuators, and the virtual body, run autonomously on a single BrainScaleS-2 microchip. The functioning network is fine-tuned for better precision and reliability through an evolution strategy. As BrainScaleS-2 emulates neural processes 1000 times faster than biology, 4800 consecutive bee journeys distributed over 320 generations occur within only half an hour on a single neuromorphic core.

What problem does this paper attempt to address?

This paper aims to solve the problem of how to use neuromorphic hardware to implement the path integration mechanism in the insect brain, so as to guide virtual bees to return to their nests in a straight line after exploring the environment. Specifically, the research team achieved this goal by simulating the neural network of the insect brain on the mixed - signal processor BrainScaleS - 2. This work not only verifies that navigation based on path integration can be implemented on accelerated neuromorphic circuits, but also shows that the neural structures derived from the insect brain can be mapped onto neuromorphic hardware to effectively control virtual agents, thereby achieving a bionic control system. ### Research Background - **Insect Navigation Ability**: Insects can navigate efficiently, quickly and robustly in complex real - world environments. This ability is partly due to the evolution of their neuroanatomical structures. - **Neurobiological Advances**: In recent years, experimental neuroscience research has made significant progress in measuring neural structures and processes, providing deep insights into their working principles. - **Path Integration Mechanism**: Research has found that the ability of bees to return to their nests in a straight line after exploring the environment stems from a path integration mechanism in the brain. ### Research Methods - **Neuromorphic Hardware**: The research team used BrainScaleS - 2, a mixed - signal neuromorphic computing system that can simulate neural dynamics 1,000 times faster than living organisms. - **Network Model**: Based on the physiological model of Megalopta genalis, the research team developed a network model of spiking - based leaky integrate - and - fire (LIF) neurons and introduced a single - cell short - term memory mechanism based on axon - axon synapses. - **Experimental Design**: The experiment was divided into two phases: the exploration phase and the return phase. In the exploration phase, virtual bees wandered randomly; in the return phase, the bees returned to their nests autonomously according to the output of the neural network. ### Main Results - **Path Integration Realization**: The experimental results show that the neuromorphic circuit can successfully implement path integration, and virtual bees can accurately return to their nests after exploration. - **Performance Optimization**: Optimizing synaptic weights through evolutionary strategies significantly improved the accuracy of the return position, with the average deviation decreasing from 4.9% to 1.0%. - **Statistical Analysis**: 1,000 independent experiments showed that the agents were always able to return in the correct direction and usually stopped at the correct distance. ### Innovation Points - **Axon - Axon Synapse**: A single - cell short - term memory mechanism based on axon - axon synapses was introduced, which was implemented on neuromorphic hardware for the first time. - **Pure Spiking Model**: A path integration model based entirely on spiking was developed, which was the first model of this kind to be published. ### Discussion - **Advantages of Neuromorphic Hardware**: Neuromorphic hardware can simulate neural dynamics at an extremely high speed, greatly shortening the experimental time. - **Application Prospects**: This research not only provides a new perspective for understanding insect navigation mechanisms, but also lays the foundation for the development of low - power, high - efficiency robot control systems. In conclusion, this paper successfully realizes the autonomous navigation of virtual bees by simulating the path integration mechanism of the insect brain on neuromorphic hardware, and further improves the navigation accuracy through optimization. This provides an important reference for future research and applications in the field of neuromorphic computing.

Emulating insect brains for neuromorphic navigation

Finding the Gap: Neuromorphic Motion Vision in Cluttered Environments

Bio-inspired spike-based Hippocampus and Posterior Parietal Cortex models for robot navigation and environment pseudo-mapping

A Memristive Spiking Neural Network Circuit for Bio-inspired Navigation Based on Spatial Cognitive Mechanisms

A neural network model for familiarity and context learning during honeybee foraging flights

Endotaxis: A neuromorphic algorithm for mapping, goal-learning, navigation, and patrolling

A neuro-cognitive robot for spatial navigation

A Navigation Cognitive System Driven by Hierarchical Spiking Neural Network.

Biomimetic FPGA-based spatial navigation model with grid cells and place cells

Toward Brain-Inspired Learning with the Neuromorphic Snake-Like Robot and the Neurorobotic Platform

A Spatial Cognition Approach Based on Grid Cell Group Representation for Embodied Intelligence

Finding the gap: neuromorphic motion-vision in dense environments

Cognitive Memory and Mapping in a Brain-Like System for Robotic Navigation

Neuromorphic implementations of neurobiological learning algorithms for spiking neural networks

Mean Field Approach for Configuring Population Dynamics on a Biohybrid Neuromorphic System

A Neuro-Cognitive System and Its Application in Robotics

An All-Neuron Spiking 2-D Path Integration and Map Representation Model Implemented on Neuromorphic Chips

Large-Scale Neuromorphic Spiking Array Processors: A Quest to Mimic the Brain

Reinforcement Learning as a Robotics-Inspired Framework for Insect Navigation: From Spatial Representations to Neural Implementation

Neuromorphic electronic circuits for building autonomous cognitive systems