xresearch is an ai research project focused on solving the most important problems in humanoids locomotion and intelligence necessary to operate autonomously.
some ideas we're interested in exploring:
robotic foundation models for full upper and lower body control
RL for human level locomotion
SoTA quantized VLAs that can run on the edge
Rubik Cube via RL then transition into other tasks that require strong reasoning skills
VLAs reasoning and complex instruction following
scaling laws for VLAs, humanoidOS, datasets, benchmarks, sim2real, tooling, etc.
more on our initial work around RL for locomotion soon!