6 papers
CAP: Codebook-Aligned Prediction for Tokenized Robot Policies
Haoran Chen, Jingtian Ji, Samuel Wheeler +2
Action tokenization converts continuous robot actions into discrete symbols that can be modeled autoregressively. However, existing tokenizer-based policies typically ignore the to…
Beyond Reconstruction: What Matters in Action Tokenization for Robot Policies?
Haoran Chen, Jingtian Ji, Samuel Wheeler +2
Autoregressive action-token policies such as vision-language-action models require action tokenizers to translate discrete token sequences into precise control actions in continuou…
What Are We Actually Benchmarking in Robot Manipulation?
Tianchong Jiang, Xiangshan Tan, Samuel Wheeler +3
A robotics benchmark score measures success under one fixed evaluation setup, yet is routinely treated as evidence of general manipulation capability. We identify four failure mode…
CLAW: Learning Continuous Latent Action World Models via Adversarial Latent Regularization
Tewodros Ayalew, Matthew Jeung, Samuel Wheeler +5
We introduce CLAW, a fully end-to-end self-supervised framework for learning a world model jointly with continuous latent action representations directly from action-free videos. O…
PROGRESSOR: A Perceptually Guided Reward Estimator with Self-Supervised Online Refinement
Tewodros Ayalew, Xiao Zhang, Kevin Yuanbo Wu +3
We present PROGRESSOR, a novel framework that learns a task-agnostic reward function from videos, enabling policy training through goal-conditioned reinforcement learning (RL) with…
Approaching Deep Learning through the Spectral Dynamics of Weights
David Yunis, Kumar Kshitij Patel, Samuel Wheeler +5
We propose an empirical approach centered on the spectral dynamics of weights -- the behavior of singular values and vectors during optimization -- to unify and clarify several phe…