Artificial Visual Perception Nervous System Based on Low-Dimensional Material Photoelectric Memristors

No prior work has fully integrated quantum-dot and nanosheet components to replicate complex biological visual pathways. That uncertainty drove the development of hardware capable of processing environmental data with high energy efficiency. It was already known that traditional computing architectures struggle to match the speed of human sensory processing. Prior research has shown that memristors offer unique advantages for mimicking synaptic behaviors in artificial intelligence. This gap motivated the exploration of low-dimensional materials for advanced photoelectric sensing applications. Scientists have long sought to bridge the divide between biological perception and synthetic hardware systems. Current electronic devices often lack the area efficiency required for sophisticated, real-time visual tasks. This study addresses these limitations by proposing a novel architecture based on memristive dynamics.

The researchers aim to develop a high-efficiency artificial visual perception nervous system using low-dimensional material memristors. This study addresses the pressing need for hardware capable of processing natural information with improved energy and area efficiency. The investigators seek to overcome limitations in current artificial intelligence systems that struggle with complex sensory tasks. They propose that memristors with elaborate dynamics offer a viable solution for mimicking biological visual pathways. The team intends to implement synaptic and leaky integrate-and-fire neuron functions through specific memristor configurations. By integrating these components, they plan to demonstrate biological image perception and sensitization processes. Furthermore, the authors aim to validate the system's utility by emulating real-world control tasks like those found in driverless automobiles. This work intends to show that hardware-based systems can systematically replicate the functions of the biological visual nervous system.

The research team designed a hardware architecture utilizing two distinct types of memristive devices. They employed quantum-dot materials to create the photoelectric component responsible for synaptic emulation. A nanosheet-based threshold-switching device was integrated to serve as the leaky integrate-and-fire neuron. The investigators tested the system by subjecting it to various visual stimuli to evaluate perception capabilities. They assessed the integration and firing responses to determine how well the hardware mimicked biological neural activity. The study also examined the biosensitization process through controlled electrical inputs. To demonstrate practical utility, the authors modeled a speed regulation task relevant to autonomous driving. This comprehensive approach allowed for the systematic evaluation of the hardware's performance against biological benchmarks.

The system successfully demonstrated biological image perception and the integration and fire process. The researchers observed that the hardware could accurately emulate the biosensitization phenomenon. Their findings show that the self-regulation of speed control in driverless automobiles is conceptually achievable with this architecture. The study confirms that the combination of quantum-dot and nanosheet memristors effectively replicates synaptic and neuronal functions. The authors report that the device processes natural information with high energy and area efficiency. These results indicate that the memristor-based hardware can systematically mirror the operations of the human visual nervous system. The data suggest that the proposed architecture expands the functional range of memristor applications in artificial intelligence. The researchers highlight that the hardware performance remains consistent across the tested biological emulation tasks.

The authors propose that their hardware architecture successfully replicates key biological visual functions. This system demonstrates the potential for memristors to emulate complex sensory processing in artificial intelligence. The researchers suggest that integrating photoelectric and threshold-switching components allows for efficient image perception. Their findings indicate that biological sensitization processes can be effectively modeled using this hardware. The team concludes that their device accurately mimics the integration and firing behaviors of neurons. This work implies that memristor-based systems could expand the current capabilities of autonomous vehicle control mechanisms. The authors note that their approach provides a scalable solution for future visual perception technologies. These results support the broader application of low-dimensional materials in neuromorphic computing architectures.