|
|
www.design-reuse-embedded.com |
Robotics of things - the next big thing in embedded
Robotics and machine learning are becoming the next big thing for embedded system hardware and software suppliers, writes Glenn Perry from Mentor Graphics.
By Richard Wilson , Dec. 12, 2016 –
Embedded systems can be characterised by connectivity and the internet of things (IoT). Whether they are smart, low energy, edge devices, intermediate gateways or compute nodes, all based on multicore system-on-chip (SoC) architectures requiring performance, reliability and security.
As a result the embedded systems industry seems to be headed into two key areas: intelligence and autonomy.
A couple areas of interest are machine learning and what I call the "Robotics of Things." In the area of machine learning, a new category of computing architectures called neuromorphic processors, or brain-inspired computing, will become mainstream.
Current Von Neumann-based architectures lack the power and performance for neural network processing, but new processing technologies, such as SpiNNaker and BrainScaleS, will mimic the brain's biological structure and behavior, capable of simulating one billion neurons in real-time.