LADA: latent actions imitate language from few observation-language pairs

abursuc · x · 2026-09-17

At the #ssad2026 workshop, the author introduces LADA (Latent Action Driving Annotations), a low-cost strategy that imitates the function of language from only a few observation-language pairs by learning to produce latent actions. The author offers an intuition for why it works: it breaks a difficult problem into two simpler ones. Details are in the linked material.

Related event: LADA: Low-Cost Language Annotations for Autonomous Driving via Latent Actions(2 posts)→

Original post →

More from Embodied

Embodied channel →