Periodic Reporting for period 1 - InsectNeuroNano (Insect-Brain inspired Neuromorphic Nanophotonics)
Reporting period: 2022-04-01 to 2023-03-31
Understanding of the neuro-architecture of key areas in the insect brain and its attached sensory systems will be used to create III-V nanowire/nanopillar and molecular dye-based network circuits that mimic neural computations underlying specific behaviours (in particular, navigation). Insights into how the sensory array of the insect eye couples to navigation control circuits will drive the development of coupled nanostructure sensor arrays and navigation systems. We will demonstrate and explore three main functionalities: connectivity, memory, and sensing; as well as concurrently develop the upscaling/commercial aspects:
• Demonstrate superior connectivity using overlapping light signals in a nanoscale system. To use light for connectivity we apply a broadcasting concept sensitizing the neural nodes to specific light signals and by sub-wavelength light manipulation of emission patterns using III-V nanowire-based components as well as molecular dyes. We will experimentally implement circuits step-by-step using a theoretically proven concept from the insect brain Central Complex. The aim to demonstrate how light for interneuron communication can have very high error tolerance and orders of magnitude better energy and spatial footprint compared to present technologies.
• Explore neuromorphic memory functionalities from nanoelectronics and molecular dyes. Based on neurobiology studies of the insect working memory we will explore how several different memory concepts can be implemented using III-V Nanowires/nanopillars and molecular dyes. The aim is to develop both an internal neural node memory, and memory in the network connections. Both short and long term memories will be explored.
• Integrate optical sensor systems and information processing. The same nanostructures used for computing will be used for optical sensing. A neural network unit will extract global orientation information from polarised skylight and time of day. The objective is to create a sensor array that output compass heading directly.
• Show upscaling, on-chip assembly and market potential. Working on scalability, energy efficiency and potential for optimization, we investigate the generality of the approach and the next steps towards commercialization. An additional outcome will be a computational approach for simulating the performance various III-V nanostructure and molecular dyes neural networks circuits.
• Demonstrate superior connectivity using overlapping light signals in a nanoscale system. To use light for connectivity we apply a broadcasting concept sensitizing the neural nodes to specific light signals and by sub-wavelength light manipulation of emission patterns using III-V nanowire-based components as well as molecular dyes. We will experimentally implement circuits step-by-step using a theoretically proven concept from the insect brain Central Complex. The aim to demonstrate how light for interneuron communication can have very high error tolerance and orders of magnitude better energy and spatial footprint compared to present technologies.
• Explore neuromorphic memory functionalities from nanoelectronics and molecular dyes. Based on neurobiology studies of the insect working memory we will explore how several different memory concepts can be implemented using III-V Nanowires/nanopillars and molecular dyes. The aim is to develop both an internal neural node memory, and memory in the network connections. Both short and long term memories will be explored.
• Integrate optical sensor systems and information processing. The same nanostructures used for computing will be used for optical sensing. A neural network unit will extract global orientation information from polarised skylight and time of day. The objective is to create a sensor array that output compass heading directly.
• Show upscaling, on-chip assembly and market potential. Working on scalability, energy efficiency and potential for optimization, we investigate the generality of the approach and the next steps towards commercialization. An additional outcome will be a computational approach for simulating the performance various III-V nanostructure and molecular dyes neural networks circuits.
For the first period of the project focus has been on connecting the highly interdisciplinary constellation tightly together and laying the groundwork for the novel light based on-chip solutions for neuromorphic computing and array sensors.
Several important milestones were reached:
1. Nanowire-to-nanowire optical on-chip communication was observed for both arrays and individual wires.
2. Dye-based memory in the optical network connections were explored by both modelling the insect navigation Central Complex neural network and synthesizing reversible memory dyes and demonstrating their function with nanowires.
3. Detailed simulations of charge-based memories in nanowire optoelectronic circuits as well as initial work on more complex nanowire device circuitry.
4. Work on exploring the polarization sensor system of insects both in biological and robotic systems as well as developing a suggestion for an efficient nanowire implementation.
Several important milestones were reached:
1. Nanowire-to-nanowire optical on-chip communication was observed for both arrays and individual wires.
2. Dye-based memory in the optical network connections were explored by both modelling the insect navigation Central Complex neural network and synthesizing reversible memory dyes and demonstrating their function with nanowires.
3. Detailed simulations of charge-based memories in nanowire optoelectronic circuits as well as initial work on more complex nanowire device circuitry.
4. Work on exploring the polarization sensor system of insects both in biological and robotic systems as well as developing a suggestion for an efficient nanowire implementation.
Three important results can be mentioned.
First, on-chip nanowire-to-nanowire optical communication in a broadcasting scheme is an important step forward to start using nanooptoelectronics in computational systems. Interestingly, while the idea of nanostructures for on-chip optical communication has been around for some times, not much has been realized for such intercommunication schemes. The optical broadcasting system technologies can also have wider implications for sensors with edge computing capabilities.
Second, an interesting part of this project is the concept of taking models and insights from insect neurobiology and applying that to neuromorphic nanophotonic hardware design. Such a tight interdisciplinary working mode is rarely seen. We have in the first part of the project demonstrated this in proposing a novel nanowire and dye hardware design for the sensory system of the insect brain that uses the polarization of the sky to form a celestial compass.
Third, we have developed a first generation of dyes with a reversible optical memory and shown that it can be applied with high efficiency on a chip operating together with semiconductor nanowire devices. Further we have found that internal memories in the nanowire circuits can improve the function of neural networks.
First, on-chip nanowire-to-nanowire optical communication in a broadcasting scheme is an important step forward to start using nanooptoelectronics in computational systems. Interestingly, while the idea of nanostructures for on-chip optical communication has been around for some times, not much has been realized for such intercommunication schemes. The optical broadcasting system technologies can also have wider implications for sensors with edge computing capabilities.
Second, an interesting part of this project is the concept of taking models and insights from insect neurobiology and applying that to neuromorphic nanophotonic hardware design. Such a tight interdisciplinary working mode is rarely seen. We have in the first part of the project demonstrated this in proposing a novel nanowire and dye hardware design for the sensory system of the insect brain that uses the polarization of the sky to form a celestial compass.
Third, we have developed a first generation of dyes with a reversible optical memory and shown that it can be applied with high efficiency on a chip operating together with semiconductor nanowire devices. Further we have found that internal memories in the nanowire circuits can improve the function of neural networks.