Open Models Almanac
Methodsother

single mention · other

55-dimensional canonical state-and-action representation

A 55-dimensional canonical vector representation encodes both states and actions for learning across multiple embodiments.1 It packs arm joints, end-effector pose, gripper, hand joints, waist, head, and mobility signals into one fixed layout.2

How this definition was written

Summarised by glm-5.3-flash from 1 source; every sentence is checked against the quote it cites.

  1. we use a 55-dimensional canonical vector representation for both states and actions — From Foundation to Application: Improving VLA Models in Practice
  2. This representation consists of 14 dimensions for arm joint position, 14 dimensions for end-effector pose, 2 dimensions for gripper position, 12 dimensions for hand joint position, 4 dimensions for waist position, 2 dimensions for head position, and 3 dimensions for mobility signal — From Foundation to Application: Improving VLA Models in Practice

Filed under other