# CT-VAM: A Cerebello-Thalamic-Inspired Vision-Action Model for Efficient Visuomotor Control

Source: [CT-VAM: A Cerebello-Thalamic-Inspired Vision-Action Model for Efficient Visuomotor Control](<https://www.sota2.com/research/paper/6a279e9590966ade919bfcff>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-10T05:15:22.625Z

Authors: Jiacheng Li, Yize Guo, Jiabin Guo, Qingchen Liu, Jiahu Qin

Published: 2026-06-08T14:46:43.000Z

## Abstract

Vision-language-action models have shown strong promise for robot manipulation, yet raw language is primarily needed to specify task intent rather than to be repeatedly processed during high-frequency low-level execution. Motivated by this separation, we propose a cerebello-thalamic-inspired vision-action model (CT-VAM) for efficient task-conditioned visuomotor control. CT-VAM acts as a compact local execution policy that predicts action chunks from dualview visual observations, proprioception, and a lightweight task condition, potentially enabling a practical cloud-edge paradigm in which high-level semantic reasoning can be handled by large models while fast closed-loop control runs on local hardware. To fuse heterogeneous inputs effectively, CT-VAM introduces TARS (Thalamic Action Routing Stream), a stream-separated conditional attention decoder that independently routes action, visual and task streams, preventing dense sensory tokens from overwhelming compact task-relevant conditions. With only 68M parameters, CT-VAM achieves LIBERO success rates competitive with substantially larger VLA models, while reducing inference latency. Together with flow-consistent inpainting for asynchronous chunk execution, CT-VAM supports high-frequency control and demonstrates robust realworld deployment on resource-constrained robotic platforms.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Ball Pouring on Real-world Ball Pouring RTX 4080](<https://www.sota2.com/research/sota/ball-pouring-on-real-world-ball-pouring-rtx-4080.md>): Ball Pouring; Real-world Ball Pouring RTX 4080; Success Rate: 100.

- [Real-world Ball Pouring Jetson Orin NX](<https://www.sota2.com/research/sota/real-world-ball-pouring-jetson-orin-nx.md>): Ball Pouring; Real-world Ball Pouring Jetson Orin NX; Success Rate: 90.

- [Visuomotor Control on LIBERO](<https://www.sota2.com/research/sota/visuomotor-control-on-libero.md>): Visuomotor Control; LIBERO; Spatial Score: 89.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2606.09572>)

[PDF](<https://arxiv.org/pdf/2606.09572.pdf>)