Is there any research on neuromorphic design for robotic interfaces?4 answersNeuromorphic design for robotic interfaces has been the focus of research. One study by Russo et al. presents an interface board and communication protocol that enables communication between different devices, using a microcontroller unit as an intermediary. Another paper by Wilmarth discusses the development of closed-loop biohybrid circuits between populations of biological neurons and neural networks of silicon neurons, which allows for the investigation of neuronal network functions and the prototyping of bidirectional neuromorphic neural interfaces. Both of these studies highlight the potential of neuromorphic engineering in creating interfaces for robotic systems.
What are the recent advances in TFT for neuromorphic applications?5 answersRecent advances in thin-film transistors (TFTs) for neuromorphic applications include the development of transistor-based neuromorphic devices. These devices, such as ion-gate neuromorphic transistors, ferroelectric-gate neuromorphic transistors, and floating-gate neuromorphic transistors, have been extensively studied and reviewed. Additionally, the use of Pb(Zr, Ti)O3 (PZT) in TFTs has shown promising results for neuromorphic systems. The high dielectric constant and polarization properties of PZT allow for reduced power consumption and a wide dynamic range in spike signals. Furthermore, the use of crystalline materials (CMs) in flexible memristors has also shown potential for neuromorphic applications. CMs-based flexible memristors, including 2D materials, metal-organic frameworks, covalent organic frameworks, and perovskites, have been extensively studied for data storage and neuromorphic devices. These recent advances in TFTs and memristors provide new opportunities for the development of energy-efficient and high-performance neuromorphic systems.
What is the influence of Neuromorphic Computing on interaction design?5 answersNeuromorphic computing has had a significant influence on interaction design. It has enabled the development of computational platforms that imitate synaptic and neuronal activities in the human brain, allowing for efficient and cognitive processing of big data flows. The advancements in neuromorphic computing hardware and system designs, particularly with non-volatile resistive access memory (ReRAM) devices, have led to high computing efficiency in terms of speed and energy. Additionally, the development of brain-inspired computing (BIC) systems has provided opportunities for the future design of neuromorphic devices, with a focus on the interactions between system, architecture, and circuit/device levels. Furthermore, the use of magnon scattering modulated by an omnidirectional mobile hopfion in antiferromagnets has allowed for the realization of a neural network with high connection density and meta-learning capabilities, breaking the connection density bottleneck in previous designs. Overall, neuromorphic computing has revolutionized interaction design by offering innovative solutions across the entire compute stack.
Are there any known implementation of neuromorphic IC in FPGA?5 answersYes, there are known implementations of neuromorphic IC in FPGA. Researchers have developed fully digital neuromorphic hardware architectures that mimic biologically inspired spiking neural networks (SNNs) using FPGA SoCs such as those proposed in the RANC framework. Field Programmable Gate Arrays (FPGAs) have been used to implement different neuron models, including the Morris-Lecar (ML) neuronal model, which has a significant impact on new designs in the field of neuromorphic engineering. Another study implemented the Izhikevich (IZH) mathematical model on an FPGA board, proposing an innovative hardware architecture for implementing a neuron. Additionally, a low-cost and SWaP (size, weight, and power) optimized neuromorphic hardware platform called μCaspian has been introduced, which uses commercial off-the-shelf components and an open-source FPGA workflow. Furthermore, an optimized hardware implementation of a neuromorphic approach for event-based object classification in FPGA has been presented.
Do neuromorphic chips have to be implemented in CMOS technology?5 answersNeuromorphic chips do not have to be implemented in CMOS technology. Different approaches have been explored for the implementation of neuromorphic chips. One approach is the use of multilevel neuromorphic logic structures based on thin film memristive compositions. Another approach is the use of thin-film semiconductor electronic devices, such as amorphous Ga-Sn-O (α-GTO) and amorphous In-Ga-Zn-O (α-IGZO) thin-film devices. Additionally, there are proposals for on-chip learning-based neuromorphic systems that perform only inference operations. Therefore, while CMOS technology has been used in some neuromorphic chip implementations, it is not the only technology that can be used. Other technologies, such as thin film memristive compositions and thin-film semiconductor electronic devices, have also been explored for the implementation of neuromorphic chips.
How the spintronics influence on the neuromorphic?5 answersSpintronics has a significant influence on neuromorphic computing. Spintronic devices, such as spintronic nanodevices and magnetic tunnel junctions, offer increased energy efficiency and decreased circuit area, making them suitable for artificial neurons and synapses in neuromorphic circuits. These devices can mimic the functionalities of biological neurons and synapses, enabling bio-mimetic computations at low terminal voltages. Spintronic resonators, particularly those based on quantum materials like transition metal oxides, can provide hysteresis and memory, enhancing the functionalities of neuromorphic systems. Antiferromagnetic spintronics also show promise for neuromorphic computing, with recent works demonstrating their potential for non-spiking and spiking neural networks. Overall, spintronics offers a bottom-up approach to overcome the limitations of conventional electronic circuits, such as high power dissipation and scaling challenges, and provides a pathway towards more efficient and brain-inspired computation in neuromorphic systems.