Abstract
Language-conditioned robot policies have made clear progress in multitask manipulation, but task-relevant local visual evidence usually stays hidden inside a visual backbone or attention layers. This leaves the policy difficult to inspect and fragile under visual change, two symptoms of a missing explicit, task-aligned local visual channel. We present TLC-DiT, a plug-in extension of the Multitask Diffusion Transformer (DiT) policy that adds explicit task-guided local visual feature maps without changing the diffusion objective or the action-generation process. For each camera view, frozen DINOv2 patch features are modulated by the CLIP task embedding through FiLM and refined by a lightweight CoordConv CNN adapter into smooth spatial maps, which are concatenated with the original global image, language, joint-state, and timestep conditions. On LIBERO, TLC-DiT reaches a 93.5% average success rate, compared with 86.5% for Multitask DiT and 79.25% for SmolVLA. On LIBERO-plus, the total success rate improves from 54.07% to 57.24%, with larger gains under camera, background, and sensor-noise changes. In real-world bimanual tasks, TLC-DiT raises Teabag Putting completion from 44% to 89% while maintaining comparable Match Box Opening performance. Feature-map visualizations confirm that the model attends to task-relevant regions across views and perturbations, providing a direct way to inspect the visual evidence.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Cai, X., Ichiwara, H., Wang, Z., Lu, Y., & Ogata, T. (2026). TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation. https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation
MLA 9
Cai, Xianbo, et al. "TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation." https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation.
Chicago (author–date)
Cai, Xianbo, Hideyuki Ichiwara, Zihang Wang, Yijun Lu, and Tetsuya Ogata. 2026. "TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation." https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation.
Harvard
Cai, X., Ichiwara, H., Wang, Z., Lu, Y. and Ogata, T. (2026) 'TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation', Available at: https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation.
Vancouver
Cai X, Ichiwara H, Wang Z, Lu Y, Ogata T. TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation. https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation
IEEE
X. Cai, H. Ichiwara, Z. Wang, Y. Lu, and T. Ogata, "TLC-DiT: Task-Aligned Local Visual Conditioning for Robust Multitask Robot Manipulation," https://omanscience.com/en/articles/tlc-dit-task-aligned-local-visual-conditioning-for-robust-multitask-robot-manipulation.