Hi Menlo!
I’m a hobbyist that’s learning robotics from the top down, having begun with the orchestrator and building down through a Unity simulation to build a generalist humanoid. We’re reaching the point where we’ll be refactoring into a fast/slow loop and moving beyond VLAs into ROS2, a world model and true simulation, with lots of learning yet to do with simpler physical forms first.
Looking at the eventual physical form to model, I arrived at the same conclusions as you did about supportability and repairability, and I really like what Menlo is building with Asimov. I’m likely a future customer, though I’m nowhere ready for a ‘here be dragons’ level.
Which leads me to some questions to the future of Asimov and about my approach. Looking ahead to dexterous hand support, do you think Menlo will be rolling your own dexterous hands or tending towards supporting existing open source designs, like the Orcas?
On power, I’ve read that the Asimov can support a Jetson Nano today, but that’s nowhere as capable as my approach would need it to be if the fast loop and the motor cortex are running locally. Will future versions upgrade the power source to support a Jetson Thor?
These are my most pressing questions for modeling in simulation and for the direction I head with compute, the latter which is a pricey one. Appreciate any guidance you can provide!