الملخص
Vision-language-action (VLA) policies assume the embodiment on which they were trained and can fail when a joint fault changes how commanded actions are physically executed. Existing fault-recovery methods often require task-specific retraining, fault labels, explicit diagnosis, or privileged embodiment information. We introduce CAPABLE, a unified capability-aware adaptation framework for frozen VLAs that integrates self-supervised capability inference with residual reinforcement learning. CAPABLE infers capability, how much of the commanded motion each joint actually realizes and how that motion contributes to end-effector behavior, online from command-response history and kinematics using a temporal encoder shared across joints, Jacobian grounding, cross-joint attention, and self-supervised physical prediction. The resulting representation conditions a residual policy that adds bounded corrections to the VLA arm action without fault labels or faulty-joint identifiers. Across 28 LIBERO tasks, CAPABLE raises success on an actuator excluded from fault training from 24.8% to 59.3%, outperforming a parameter-matched global-history baseline by 17.4 points while preserving healthy performance. Leave-one-actuator-out experiments across six joints show that this transfer is not specific to one actuator, and additional evaluations characterize transfer to unseen fault families and demonstrate recovery on a physical Franka Panda. https://capable-vla.github.io/
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Khoshnazar, M., Tezerjani, M. D., Qu, D., Gao, Z., Zhan, Y., Schafer, J., Melnik, A., Yang, Q., & Beetz, M. (2026). CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding. https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding
MLA 9
Khoshnazar, Mohammad, et al. "CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding." https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding.
شيكاغو (المؤلف–التاريخ)
Khoshnazar, Mohammad, Mohammad Dehghani Tezerjani, Deyuan Qu, Zhiyuan Gao, Yanxiang Zhan, Jeroen Schafer, Andrew Melnik, Qing Yang, and Michael Beetz. 2026. "CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding." https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding.
هارفارد
Khoshnazar, M., Tezerjani, M. D., Qu, D., Gao, Z., Zhan, Y., Schafer, J., Melnik, A., Yang, Q. and Beetz, M. (2026) 'CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding', Available at: https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding.
فانكوفر
Khoshnazar M, Tezerjani MD, Qu D, Gao Z, Zhan Y, Schafer J, et al. CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding. https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding
IEEE
M. Khoshnazar, M. D. Tezerjani, D. Qu, Z. Gao, Y. Zhan, J. Schafer, A. Melnik, Q. Yang, and M. Beetz, "CAPABLE: Capability-Aware Policy Adaptation via Behavioral Latent Encoding," https://omanscience.com/ar/articles/capable-capability-aware-policy-adaptation-via-behavioral-latent-encoding.