Organizations: University of Science and Technology of China · Tongji University · The Chinese University of Hong Kong · Anhui University · Nanjing University
In the real world, mobile manipulation robots must perform complex object interactions and make multi-step decisions, often relying on task-relevant information that is no longer observable. However, existing benchmarks provide limited evaluation of such memory-dependent decision-making in mobile manipulation. To address this gap, we introduce RoboCasa-Mem, a new benchmark comprising nine mobile manipulation tasks that require agents to use interaction history at key decision points. Specifically, the tasks are grouped into two categories of memory: One is environment memory, which captures spatial locations, object attributes, and object relationships. The other is process memory, which captures single-task progress and multi-task scheduling. Particularly, we introduce Behavior-Conditioned Evaluation. For tasks with multiple valid execution paths, instead of fixing a single target or trajectory, the evaluator derives later task constraints from the robot's earlier actions. This allows us to evaluate whether later behavior is consistent with the robot's own interaction history while reducing shortcuts based on fixed target assignments. We provide 1,800 successful demonstrations for policy training. We also develop a baseline on our benchmark, namely Hierarchical Memory-Augmented Policy(HMAP). Its high-level planner uses structured entity and process memories to generate subtasks. Its low-level VLA controller uses subgoal-conditioned queries to incorporate VGGT-Omega geometry into action generation. Together, RoboCasa-Mem and HMAP baseline support systematic evaluation of memory-dependent behavior and the development of memory-aware mobile manipulation policies.
Figures & tables
Memory coverage
Benchmark
Mobile
History [-1pt]dependence
Spatial
Attribute
Relation
Progress
Scheduling
Stage [-1pt]eval
MIKASA-Robo [-2pt] ( Cherepanov et al., 2025 )
✗
✓
✓
✗
✓
✓
✗
✗
RMBench [-2pt] ( Chen et al., 2026 )
✗
✓
✓
✗
✓
✓
✗
✗
RoboMME [-2pt] ( Dai et al., 2026 )
✗
✓
✓
✗
✓
✓
✗
✗
RoboTwin-Mem [-2pt] ( Yang et al., 2026 )
✗
✓
✓
✗
✓
✓
✗
✗
RoboMemArena [-2pt] ( Lei et al., 2026 )
✗
✓
✓
✓
✓
✓
✗
✓
Table 1: Comparison of simulated robotic memory benchmarks. Coverage denotes at least one qualifying task, not uniform requirements across tasks or equivalent taxonomies.
Figure 1: Overview of RoboCasa-Mem. Top: a history-dependent cup-selection example illustrates why the current observation alone may not determine the correct action. Middle: five memory categories cover spatial, attribute, and relational Environment Memory, together with single-task progress and multi-task scheduling in Process Memory. Bottom: demonstration construction and semantic evaluation support analysis of final outcomes and stage-level completion. Table 2 lists the nine core tasks.
Task
Memory type
Key challenge
Description
Environment Memory
PutBackCup
Spatial
Location recall
“Place the cup at the previously cleaned spot.”
SweetCup
Attribute
Hidden attribute
“Prepare one sweetened and one unsweetened coffee, then select the sweetened cup.”
TempReport
Attribute
Attribute recall
“Recall whether hot or cold water was dispensed, then press the matching button.”
CanRecall
Relational
Object–container binding
“After closing the cabinets, recall the requested items’ locations and place them back on the tabletop.”
MilkMatch
Relational
Source–target binding
“Transport milk boxes in two trips, placing each on the plate matching its source tray’s color.”
Table 2: Overview of the nine RoboCasa-Mem tasks, organized into five memory categories under Environment Memory and Process Memory. Quoted descriptions summarize the tasks rather than reproduce the exact policy instructions.
Figure 2: Overview of our Hierarchical Memory-Augmented Policy. The high-level planner maintains entity and process memory to select an active subgoal. The low-level subgoal-guided 3D-aware controller conditions on this subgoal and retrieves task-relevant features from current multi-view geometry to modulate action generation.
(a) Environment Memory
Model
PutBackCup
SweetCup
TempReport
CanRecall
MilkMatch
SR
SCR
SR
SCR
SR
SCR
SR
SCR
SR
SCR
ACT
0.00
0.48
0.00
0.07
0.00
0.09
0.00
0.09
0.00
0.03
DP
0.00
0.20
0.00
0.04
0.08
0.21
0.00
0.05
0.00
0.05
π0.5
0.68
0.85
0.04
0.27
0.22
0.57
0.00
0.24
0.12
0.47
MemER-style
0.10
0.22
0.00
0.21
0.00
0.58
0.00
0.21
0.00
0.19
Table 3: Results on the nine RoboCasa-Mem tasks. SR denotes task success rate, and SCR denotes stage completion rate. Avg. is the macro-average across all nine tasks. Best nonzero values are bolded, including ties at the reported precision.
Figure 3: An illustration of our real-world experiments conducted on AgileX Cobot Magic Platform. The task is set as: the robot first observes and picks up an object, then turns to transport it and places it into the corresponding container.
Table 7
Appendix figures & tables3 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 4: Execution workflows of the nine RoboCasa-Mem tasks.
Method
Clean
Tool
Return
ACT
0.72
0.72
0.00
DP
0.30
0.30
0.00
π0.5
0.94
0.94
0.68
MemER-style
0.28
0.28
0.10
HMAP (Ours)
0.84
0.84
0.70
Appendix
Table 6: Stage-level completion rates for PutBackCup, SweetCup, and TempReport. Each value is the fraction of evaluated episodes satisfying the corresponding stage and its required predecessors.
Method
Close-1
Close-2
Reopen
Success
ACT
0.34
0.00
0.00
0.00
DP
0.18
0.00
0.00
0.00
π0.5
0.76
0.14
0.04
0.00
MemER-style
0.80
0.02
0.00
0.00
HMAP (Ours)
0.80
0.24
0.00
0.00
Appendix
Table 7: Stage-level completion rates for the remaining six RoboCasa-Mem tasks. Abbreviated stage names are explained below each subtable.
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/.
Robotic manipulation often requires acting on information that is no longer visible, yet Vision-Language-Action policies are usually evaluated when the current observation largely determines the next action. Existing robotic memory benchmarks expose this gap, but they still rely mainly on final task success and therefore conflate forgetting with manipulation failure. We present MEMOBench, a benchmark for process level memory evaluation in robotic manipulation. MEMOBench includes 30 history dependent tasks, 1{,}500 expert demonstrations, and 4{,}200 executable checkpoint instances from 84 templates. Each checkpoint pairs coarse to fine language with a simulator predicate and labels one memory operation: Storage, Update, or Compression. These annotations define Memory Storage Rate, Memory Update Rate, and Memory Compression Rate, which measure memory fidelity alongside task success. Across standard and memory augmented VLA policies, the strongest memory module baseline reaches only 31.9% average success rate, and high storage often coexists with weak update and compression. Checkpoint language also supervises semantic, contrastive, and framewise memory alignment objectives, yielding modest gains across different memory operations. MEMOBench provides a diagnostic evaluation suite and training supervision for memory grounded robotic policies. The project page is available at https://github.com/Collab-Gen/MEMOBench.
Memory is critical for long-horizon and history-dependent robotic manipulation. Such tasks often involve counting repeated actions or manipulating objects that become temporarily occluded. Recent vision-language-action (VLA) models have begun to incorporate memory mechanisms; however, their evaluations remain confined to narrow, non-standardized settings. This limits systematic understanding, comparison, and progress measurement. To address these challenges, we introduce RoboMME: a large-scale standardized benchmark for evaluating and advancing VLA models in long-horizon, history-dependent scenarios. Our benchmark comprises 16 manipulation tasks constructed under a carefully designed taxonomy that evaluates temporal, spatial, object, and procedural memory. We further develop a suite of 14 memory-augmented VLA variants built on the π0.5 backbone to systematically explore different memory representations across multiple integration strategies. Experimental results show that the effectiveness of memory representations is highly task-dependent, with each design offering distinct advantages and limitations across different tasks. Videos and code can be found at our website https://robomme.github.io.
Yinpei Dai, Hongze Fu, Jayjun Lee +6
University of Michigan · Stanford University · Figure AI