Reference Number: ARL-SRP-00053
Project Description
Objective: The objective of this project/opportunity is to implement neuromorphic circuits on FPGA board using Verilog to demonstrate Army-relevant applications, and to create a framework or modular code to create neuromorphic circuits from core building blocks (e.g. synapses, neurons, address-event representation, encoding/decoding blocks, routing fabric, timing/delay buffers, I/O interfaces).
Purpose: Provide a method to rapidly demonstrate ferroelectric-based neuromorphic circuits for Army edge computing platforms to test different options/configurations before fabrication or tape-out. Future Army platforms need to process massive amounts of sensor data in real-time under power and connectivity constraints. This project helps enable rapid prototyping of low-power, edge computing system components.
Proposed Work: Develop Verilog code modules to emulate functionality of specified neuromorphic building blocks, such as neurons, synapses, input/output encoders, and timing buffers. Then implement these building blocks on FPGA boards using AMD Vivado software suite to construct neuromorphic circuits suitable for use with spiking neural networks, connecting them with address-even representation (AER). Finally, interface the FPGA board with appropriate function generators or computers for input and with appropriate sensors/meters or computers for output.
Requisite Tool Skills: Verilog programming, AMD Vivado software suite, FPGA hardware programming, source-meters, and function generators
Desired Knowledge: Digital logic design, basic understanding of neurons and synapses, along with some exposure to neural networks
Deliverables: A functional demonstration of a specified neuromorphic circuit (e.g. ferroelectric crossbar array with peripheral and readout circuits) that can accept a specified input and provide an expected output.
Technical Skills
- Artificial Intelligence/Machine Learning
- Computer Science
- Electrical Engineering
- Electronics
- AMD Vivado
- Verilog
- Python
