Search for: All records

Miniature robots are useful during disaster response and accessing remote or unsafe areas. They need to navigate uneven terrains without supervision and under severe resource constraints such as limited compute, storage and power budget. Event-based sensorimotor control in edge robotics has potential to enable fully autonomous and adaptive robot navigation systems capable of responding to environmental fluctuations by learning new types of motion and real-time decision making to avoid obstacles. This work presents a novel bio-inspired framework with a hierarchical control system to address these limitations, utilizing a tunable multi-layer neural network with a hardware-friendly Central Pattern Generator (CPG) as the core coordinator to govern the precise timing of periodic motion. Autonomous operation is managed by a Dynamic State Machine (DSM) at the top of the hierarchy, providing the necessary adaptability to handle environmental challenges such as obstacles or uneven terrain. The multi-layer neural network uses a nonlinear neuron model which employs mixed feedback at multiple timescales to produce rhythmic patterns of bursting events to control the motors. A comprehensive study of the architecture's building blocks is presented along with a detailed analysis of network equations. Finally, we demonstrate the proposed framework on the Petoi robot, which can autonomously learn walk and crawl gaits using supervised Spike-Time Dependent Plasticity (STDP) learning algorithm, transition between the learned gaits stored as new states, through the DSM for real-time obstacle avoidance. Measured results of the system performance are summarized and compared with other works to highlight our unique contributions. 

Abstract Neuromorphic computing shows promise for advancing computing efficiency and capabilities of AI applications using brain-inspired principles. However, the neuromorphic research field currently lacks standardized benchmarks, making it difficult to accurately measure technological advancements, compare performance with conventional methods, and identify promising future research directions. This article presents NeuroBench, a benchmark framework for neuromorphic algorithms and systems, which is collaboratively designed from an open community of researchers across industry and academia. NeuroBench introduces a common set of tools and systematic methodology for inclusive benchmark measurement, delivering an objective reference framework for quantifying neuromorphic approaches in both hardware-independent and hardware-dependent settings. For latest project updates, visit the project website (neurobench.ai). 

Abstract Deoxyribonucleic acid (DNA) has emerged as a promising building block for next-generation ultra-high density storage devices. Although DNA has high durability and extremely high density in nature, its potential as the basis of storage devices is currently hindered by limitations such as expensive and complex fabrication processes and time-consuming read–write operations. In this article, we propose the use of a DNA crossbar array architecture for an electrically readable read-only memory (DNA-ROM). While information can be ‘written’ error-free to a DNA-ROM array using appropriate sequence encodings its read accuracy can be affected by several factors such as array size, interconnect resistance, and Fermi energy deviations from HOMO levels of DNA strands employed in the crossbar. We study the impact of array size and interconnect resistance on the bit error rate of a DNA-ROM array through extensive Monte Carlo simulations. We have also analyzed the performance of our proposed DNA crossbar array for an image storage application, as a function of array size and interconnect resistance. While we expect that future advances in bioengineering and materials science will address some of the fabrication challenges associated with DNA crossbar arrays, we believe that the comprehensive body of results we present in this paper establishes the technical viability of DNA crossbar arrays as low power, high-density storage devices. Finally, our analysis of array performance vis-à-vis interconnect resistance should provide valuable insights into aspects of the fabrication process such as proper choice of interconnects necessary for ensuring high read accuracies. 

Deoxyribonucleic acid (DNA) has emerged as a promising building block for designing next-generation ultra-high density storage devices. Although DNA is highly durable and extremely high density in nature, its potential as the basis of storage devices is currently hindered by limitations such as expensive and complex fabrication processes and time-consuming read-write operations. In this article, we propose the use of a DNA crossbar array architecture for an electrically-readable Read-Only Memory (DNA-ROM). For DNA-ROM, we have chosen two DNA strands for representing Bit 1 and Bit 0 respectively. DNA charge transport has been studied through a contact-DNA-contact setup. The results obtained from the DNA charge transport study have been used to analyze the crossbar array. The performance has been analyzed by loading an image onto a 128×128 crossbar. For this application, we have observed a bit error rate of 4.52% and power consumption of 6.75 µW.