RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917344139280384 |
|---|---|
| author | Chen, Tianxing Wang, Yuran Li, Mingleyang Qin, Yan Shi, Hao Li, Zixuan Hu, Yifan Zhang, Yingsheng Wang, Kaixuan Chen, Yue Wang, Hongcheng Xu, Renjing Wu, Ruihai Mu, Yao Yang, Yaodong Dong, Hao Luo, Ping |
| author_facet | Chen, Tianxing Wang, Yuran Li, Mingleyang Qin, Yan Shi, Hao Li, Zixuan Hu, Yifan Zhang, Yingsheng Wang, Kaixuan Chen, Yue Wang, Hongcheng Xu, Renjing Wu, Ruihai Mu, Yao Yang, Yaodong Dong, Hao Luo, Ping |
| contents | Robotic manipulation policies have made rapid progress in recent years, yet most existing approaches give limited consideration to memory capabilities. Consequently, they struggle to solve tasks that require reasoning over historical observations and maintaining task-relevant information over time, which are common requirements in real-world manipulation scenarios. Although several memory-aware policies have been proposed, systematic evaluation of memory-dependent manipulation remains underexplored, and the relationship between architectural design choices and memory performance is still not well understood. To address this gap, we introduce RMBench, a simulation benchmark comprising 9 manipulation tasks that span multiple levels of memory complexity, enabling systematic evaluation of policy memory capabilities. We further propose Mem-0, a modular manipulation policy with explicit memory components designed to support controlled ablation studies. Through extensive simulation and real-world experiments, we identify memory-related limitations in existing policies and provide empirical insights into how architectural design choices influence memory performance. The website is available at https://rmbench.github.io/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_01229 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design Chen, Tianxing Wang, Yuran Li, Mingleyang Qin, Yan Shi, Hao Li, Zixuan Hu, Yifan Zhang, Yingsheng Wang, Kaixuan Chen, Yue Wang, Hongcheng Xu, Renjing Wu, Ruihai Mu, Yao Yang, Yaodong Dong, Hao Luo, Ping Robotics Artificial Intelligence Robotic manipulation policies have made rapid progress in recent years, yet most existing approaches give limited consideration to memory capabilities. Consequently, they struggle to solve tasks that require reasoning over historical observations and maintaining task-relevant information over time, which are common requirements in real-world manipulation scenarios. Although several memory-aware policies have been proposed, systematic evaluation of memory-dependent manipulation remains underexplored, and the relationship between architectural design choices and memory performance is still not well understood. To address this gap, we introduce RMBench, a simulation benchmark comprising 9 manipulation tasks that span multiple levels of memory complexity, enabling systematic evaluation of policy memory capabilities. We further propose Mem-0, a modular manipulation policy with explicit memory components designed to support controlled ablation studies. Through extensive simulation and real-world experiments, we identify memory-related limitations in existing policies and provide empirical insights into how architectural design choices influence memory performance. The website is available at https://rmbench.github.io/. |
| title | RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2603.01229 |