Autoware Vision Pilot on Axelera

Question

This project goes one level deeper than the typical AI pipeline flow, where several pre-compiled models are assembled into an application.

The current Autoware Perception stack (Vision Pilot : https://github.com/autowarefoundation/vision_pilot) was designed for NVIDIA, and does not compile for Axelera Metis.

This is today’s reality, a situation that engineers often face when moving an algorithm from NVIDIA to more efficient embedded hardware.

This project will explore the use of Wingman to identify the problematic unsupported layers, modify the model architecture, and re-train the models for Axelera Metis.

If successful, this will be ground-breaking … an Edge AI NPU that can port a model developed for NVIDIA to itself !

If not successful, this will be a good use case for further exploration by the Axelera innovation team.

Cheers !

Mario Bergeron (A.K.A. AlbertaBeef)

Spanner · Accepted Answer

That’d be a real Wingman right there!🤣 If it doesn’t currently work for the hardware, itmakesit work, and then helps you run it even better!I think there’s a similar feature in Wingman for porting from other platforms, but I’d love to see this in operation too, given a really robust and solid foundation for the application in question.Another exciting non-vision project! Great work.