Why it matters
Robot learning work often stalls between simulation and real hardware. A room of builders with working benches lets you get a second pair of eyes on a policy, a sim-to-real gap or a bug.
The tokenmaxxing angle
Ask how much inference a robot policy needs per second of operation, and whether a small onboard model with occasional calls to a larger remote one beats calling a big model on every step.
From the organizers
Mon Nov 23, 6-9 PM PT at Mission Robotics, Mission District, SF. Host Mattie Fairchild. Free, no approval required, 84 guests registered. Round of intros at 6, open build time 6:30 to 8:30, then a second round of show-and-tell.