Brain-inspired analogue systems to boost the efficiency of AI drones
A grant to develop new analogue hardware, with potential applications toward more efficient AI-based drones, has been awarded to engineers at the University of Rochester.
The AI systems that guide drones and self-driving cars rely on neural networks inspired by the human brain. However, the digital computers they run on, though ultra-reliable, suffer from high power consumption because they were initially designed for general-purpose computing tasks. Analogue circuits are more suitable in miicing the brain.
To develop the new energy-efficient analogue hardware, the engineers are abandoning conventional state-of-the-art neural networks, typically built on digital hardware for computer vision, and instead turning to predictive coding networks. Such networks are based on neuroscience theories that the brain has a mental model of the environment and constantly updates it based on feedback from the eyes.
“Research by neuroscientists has shown that the workhorse of developing neural networks — this mechanism called back propagation — is biologically implausible and our brains’ perception systems don’t work that way,” says Michael Huang, a professor of electrical and computer engineering, computer science, data science and artificial intelligence at Rochester. “To solve the problem, we asked how our brains do it. The prevailing theory is predictive coding, which involves a hierarchical process of prediction and correction — think paraphrasing what you heard, telling it to the speaker, and using their feedback to refine your understanding.”
The Rochester-led team includes Huang and electrical and computer engineering professors Hui Wu and Tong Geng, as well as their students, along with two research groups from Rice University and UCLA. The team will receive up to $7.2 million from the Defence Advanced Research Projects Agency (DARPA) over the next 54 months to develop biologically inspired predictive coding networks for digital image recognition built on analogue circuits.
While the initial prototype will focus on classifying static images, the engineers aim to bring the analogue system up to the performance level of existing digital approaches. Achieving this milestone will enable the analogue system to be applied to more complex perception tasks required by self-driving cars and autonomous drones.
And while the approach is novel, the system will not use any experimental devices but will instead be manufactured using existing technologies like the complementary metal oxide semiconductor (CMOS).
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
