Accessible event-address processing: lowering the barrier for neuromorphic tech

Neuromorphic sensors typically emit address-event representations (AER), but integrating such sensors with our infrastructure is difficult and time consuming. We all feel this pain, and we can do much better as a community.

In this discussion session, we'll present our take on how we can lower the barrier-to-entry for working with neuromorphic sensors and processors. We'll briefly demo our work on AEStream [1], but the main focus will be on how we, as a community, can make neuromorphic AER technologies more open and accessible. Ideally, by finding a way to collaborate on shared software/documentation/other things that

• Let's us plug-and-play sensors within minutes, not hours


• Integrates with modern frameworks like Numpy, PyTorch, Jax, etc.


• Shares tools and methods for debugging/visualization/inspection


• Supports conventional and, sometimes, resource-constrained computers


• Works in real-time

Perhaps you have similar code/software to share? Reach out at jeped@kth.se

Looking forward to our discussion!

[1]: https://github.com/aestream/aestream

Go to group wiki Go to wiki users Info

Timetable

No timetable published yet.

Moderator

Member