Adaptive Robotic Arm Control with a Spiking Recurrent Neural Network on a Digital Accelerator
This work addresses the problem of efficient neural network deployment in embedded systems for robotics, though it is incremental as it adapts an existing chip to a new platform and application.
The researchers tackled the challenge of deploying spiking neural networks for robotic arm control by adapting the ReckOn chip to a Xilinx MPSoC system, achieving a peak performance of 3.8M events per second while preserving simulated accuracy.
With the rise of artificial intelligence, neural network simulations of biological neuron models are being explored to reduce the footprint of learning and inference in resource-constrained task scenarios. A mainstream type of such networks are spiking neural networks (SNNs) based on simplified Integrate and Fire models for which several hardware accelerators have emerged. Among them, the ReckOn chip was introduced as a recurrent SNN allowing for both online training and execution of tasks based on arbitrary sensory modalities, demonstrated for vision, audition, and navigation. As a fully digital and open-source chip, we adapted ReckOn to be implemented on a Xilinx Multiprocessor System on Chip system (MPSoC), facilitating its deployment in embedded systems and increasing the setup flexibility. We present an overview of the system, and a Python framework to use it on a Pynq ZU platform. We validate the architecture and implementation in the new scenario of robotic arm control, and show how the simulated accuracy is preserved with a peak performance of 3.8M events processed per second.