Time-Frequency Geometric Cross-Attention for Chunked Vision-Language-Action Models

• URL: http://arxiv.org/abs/2609.09925v1
• Date: Wed, 09 Sep 2026 09:16:49 GMT
• Title: Time-Frequency Geometric Cross-Attention for Chunked Vision-Language-Action Models
• Abstract summary: We show how to repair two blind spots in a vision--action (VLA) policy.<n>We use a drop-in module to fine-tune a pretrained VLA.<n>We show how TFGCA improves in-distribution LIBERO by +1.5 on average, the OOD LIBERO-Plus by +6.3, the randomized average under RoboTwin domain randomization by +28.5, and the overall success rate on three real-robot AgiBot A2 tasks by +11.67 points.
• Score: 10.591519354017803
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Modern vision-language-action (VLA) policies predict a whole chunk of actions: one to two seconds of coordinated motion emitted in a single forward pass. Yet an action chunk is essentially a short multivariate trajectory, but inside these models it is a sequence of generic per-timestep hidden tokens decoded by a linear head. This under-serves two motion structures. First, frequency: a chunk superimposes a smooth global trend and fine corrective motion across time scales, and a single token entangles them. Second, cross-phase geometry: motions of different phases (reach, contact, grasp adjustment, settling) unfold along very different, near-orthogonal directions in representation space, yet are tightly related for the task and arise across the time axis. Dot-product attention scores alignment by an inner product, so it favors aligned tokens and is least sensitive near orthogonality, leaving such relationships for the network to recover through a detour. We introduce Time-Frequency Geometric Cross-Attention (TFGCA), a drop-in module repairing both blind spots. TFGCA uses a per-dimension learnable stationary wavelet transform to decompose the action chunk into time-frequency tokens, and each time token retrieves information from them via a cross-attention that fuses the dot product (similarity) with the wedge-product magnitude (sensitive to near-orthogonality) through a learnable weight. A zero-initialized residual reproduces the base behavior at initialization, so it can be dropped onto a pretrained VLA and fine-tuned jointly. Relative to the same-source base, TFGCA improves in-distribution LIBERO by +1.5 on average, the OOD LIBERO-Plus by +6.3, the randomized average under RoboTwin domain randomization by +28.5, and the overall success rate on three real-robot AgiBot A2 tasks by +11.67 points, with larger gains out of distribution.

Related papers

• A*-Thought-V2: Efficient Latent Reasoning via Geometric Dynamics of LLM [40.07171947823021]
  Chain-of-Thought (CoT) improves the reasoning ability of Large Language Models (LLMs) but incurs substantial computation and context costs.<n>We present A*-Thought-V2, a geometric dynamics of LLM guided framework that models CoT as a hidden-state trajectory and replaces hard deletion with an explicit-implicit interleaved latent architecture.
  arXiv  Detail & Related papers  (2026-09-07T17:56:20Z)
• The Quantum Shortcut: Complex Phase-State Dynamics Reduce the Optimization Steps of Sequence Models [0.0]
  Sequence models are distinguished by their backbone, the mechanism that routes information across positions.<n>This paper varies a choice that is prior to the backbone and shared by nearly all current models.<n>We study a complex-valued alternative drawn from the mathematics of quantum theory, in which information is carried by the phases of the state.
  arXiv  Detail & Related papers  (2026-08-07T15:12:36Z)
• TIE: Time Interval Encoding for Video Generation over Events [50.66585165263848]
  Director-style prompting, robotic action prediction, and interactive video agents demand temporal grounding over concurrent events.<n>Modern video generators represent time as discrete points through point-wise positional encodings.<n>Time Interval TIE is a principled, plug-and-play interval-aware generalization.
  arXiv  Detail & Related papers  (2026-05-11T13:23:14Z)
• MoCapAnything V2: End-to-End Motion Capture for Arbitrary Skeletons [56.68975315643491]
  We present the first fully end-to-end framework in which Videoto-Pose and Pose-to-Rotation are learnable and optimized jointly.<n>We show that our method reduces rotation error from 17 degrees to 10 degrees, and to 6.54 degrees on unseen skeletons, while achieving 20x faster inference than mesh-based pipelines.
  arXiv  Detail & Related papers  (2026-04-30T17:16:38Z)
• CAWN: Continuous Acoustic Wave Networks for Autoregressive Language Modeling [46.16066322190728]
  We introduce the Continuous Acoustic Wave Network (CAWN), a fully continuous sequence-mixing architecture.<n>Instead of discrete matrix-based attention, CAWN projects hidden states into multi-headed complex-domain phasors.<n>To prevent signal degradation over ultra-long contexts, we introduce a dual-gated Selective Phase Resonance mechanism.
  arXiv  Detail & Related papers  (2026-04-05T20:13:22Z)
• OmniSAT: Compact Action Token, Faster Auto Regression [70.70037017501357]
  We introduce an Omni Swift Action Tokenizer, which learns a compact, transferable action representation.<n>The resulting discrete tokenization shortens the training sequence by 6.8$times$, and lowers the target entropy.
  arXiv  Detail & Related papers  (2025-10-08T03:55:24Z)
• Leveraging 2D Information for Long-term Time Series Forecasting with Vanilla Transformers [55.475142494272724]
  Time series prediction is crucial for understanding and forecasting complex dynamics in various domains. We introduce GridTST, a model that combines the benefits of two approaches using innovative multi-directional attentions. The model consistently delivers state-of-the-art performance across various real-world datasets.
  arXiv  Detail & Related papers  (2024-05-22T16:41:21Z)
• Coordinate Transformer: Achieving Single-stage Multi-person Mesh Recovery from Videos [91.44553585470688]
  Multi-person 3D mesh recovery from videos is a critical first step towards automatic perception of group behavior in virtual reality, physical therapy and beyond. We propose the Coordinate transFormer (CoordFormer) that directly models multi-person spatial-temporal relations and simultaneously performs multi-mesh recovery in an end-to-end manner. Experiments on the 3DPW dataset demonstrate that CoordFormer significantly improves the state-of-the-art, outperforming the previously best results by 4.2%, 8.8% and 4.7% according to the MPJPE, PAMPJPE, and PVE metrics, respectively.
  arXiv  Detail & Related papers  (2023-08-20T18:23:07Z)
• Prune Spatio-temporal Tokens by Semantic-aware Temporal Accumulation [89.88214896713846]
  STA score considers two critical factors: temporal redundancy and semantic importance. We apply the STA module to off-the-shelf video Transformers and Videowins. Results: Kinetics-400 and Something-Something V2 achieve 30% overshelf reduction with a negligible 0.2% accuracy drop.
  arXiv  Detail & Related papers  (2023-08-08T19:38:15Z)
• Low-Rank Autoregressive Tensor Completion for Multivariate Time Series Forecasting [15.927264813606637]
  Time series collected from sensor networks are often large-scale and incomplete with considerable corruption and missing values. We propose a low-rank autoregressive tensor completion (LATC) framework to model multivariate time series data. Our numerical experiments on three real-world data sets demonstrate the superiority of the integration of global and local trends in LATC.
  arXiv  Detail & Related papers  (2020-06-18T11:31:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.