Evaluation methodologies and benchmarks.
Yandong Wang, Jiaqian Yu, Xiongfeng Peng et al.
Vision-language-action (VLA) models show strong capabilities in single and dual-arm robotic manipulation. Prior works show coordinated bimanual behaviors can emerge from end-to-end learning, leveraging large vision-language backbones with continuous action prediction.
Yandong Wang, Jiaqian Yu, Xiongfeng Peng et al.
Vision-language-action (VLA) models show strong capabilities in single and dual-arm robotic manipulation. Prior works show coordinated bimanual behaviors can emerge from end-to-end learning, leveraging large vision-language backbones with continuous action prediction.
Gia-Binh Nguyen, Trong-Bao Ho, Thien-Loc Ha et al.
Vision-Language-Action (VLA) models pre-trained on massive video-robot datasets have revolutionized robotic manipulation, yet their multi-billion parameter architectures impose prohibitive computational burdens during downstream fine-tuning and real-time inference.
Gia-Binh Nguyen, Trong-Bao Ho, Thien-Loc Ha et al.
Vision-Language-Action (VLA) models pre-trained on massive video-robot datasets have revolutionized robotic manipulation, yet their multi-billion parameter architectures impose prohibitive computational burdens during downstream fine-tuning and real-time inference.
Jianing Guo, Fangzheng Chen, Zihao Mao et al.
Flow matching has emerged as a standard paradigm for robotic manipulation owing to its strong expressive power for modelling complex, multimodal action distributions, alongside similar approaches like diffusion policy.
Jianing Guo, Fangzheng Chen, Zihao Mao et al.
Flow matching has emerged as a standard paradigm for robotic manipulation owing to its strong expressive power for modelling complex, multimodal action distributions, alongside similar approaches like diffusion policy.
Chenyang Ma, Yue Yang, Radu Corcodel et al.
FurnitureVLA makes assembly progress an explicit prediction, helping a bimanual policy decide when to move to the next subtask.
Yu Zhang, Kangyi Ji, Yongxiang Zou et al.
This paper presents an invertible neural network adapter for general robotic manipulation, designed to generate precise high-dimensional actions conditioned on multimodal observations, including visual, linguistic, and proprioceptive inputs, through a one-step denoising process.
Yu Zhang, Kangyi Ji, Yongxiang Zou et al.
This paper presents an invertible neural network adapter for general robotic manipulation, designed to generate precise high-dimensional actions conditioned on multimodal observations, including visual, linguistic, and proprioceptive inputs, through a one-step denoising process.
William English, Hao Zheng, Rickard Ewetz
This method checks a predicted action sequence for future collisions while the VLA is still generating it, rather than correcting only the next action.
Arnav Balaji, Arpit Bahety, Sriniket Ambatipudi et al.
While robotic manipulation capabilities have advanced rapidly, physical safety remains a major barrier to deploying household robots: task success is insufficient if the robot damages itself or its surroundings.
Sergey Arkhangelskiy
Real-world evaluation of vision-language-action (VLA) policies still rests on binary success rate at a fixed timeout with N <= 25 rollouts per condition, almost always without confidence intervals or
Read these papers in order to build expertise in evaluation.
Yandong Wang, Jiaqian Yu et al. · 2026
Yandong Wang, Jiaqian Yu et al. · 2026
Gia-Binh Nguyen, Trong-Bao Ho et al. · 2026
Gia-Binh Nguyen, Trong-Bao Ho et al. · 2026
Jianing Guo, Fangzheng Chen et al. · 2026
Jianing Guo, Fangzheng Chen et al. · 2026
…and 6 more papers.
