2AM: Grounding Agent-Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation

• URL: http://arxiv.org/abs/2609.11308v1
• Date: Thu, 10 Sep 2026 09:35:56 GMT
• Title: 2AM: Grounding Agent-Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation
• Abstract summary: Long-horizon robot manipulation requires memory, but not necessarily inside the action policy.<n>2AM makes a multimodal Agent the sole holder of task memory and a single RGB-based, episodically stateless Action Model the sole executor of task-relevant motion.<n>On LIBERO-Mem, without depth, online geometry, or planner-based object motion, 2AM reaches 76.3% average completion, a 61.5-point improvement over the strongest reported baseline of 14.8%, together with 63.0% relaxed and 11.8% strict success.
• Score: 6.040186174770221
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Long-horizon robot manipulation requires memory, but not necessarily inside the action policy. To address such tasks, current agentic systems often combine VLAs with planners and geometric tools, sometimes using additional depth or calibrated geometry. These systems confound attribution: gains may come from richer observations or alternative motor tools, while failures may stem from either the policy or an under-specified language interface. We isolate this question through a deliberately constrained design: less tool breadth, but greater interface bandwidth. 2AM makes a multimodal Agent the sole holder of task memory and a single RGB-based, episodically stateless Action Model the sole executor of task-relevant motion. The Agent compiles interaction history into subtask language and optional 2D grasp, place, and move hints that bind its physical intention at different time scales. To teach this steerability to the VLA, we augment demonstrations with structured hint labels and train under condition dropout, spatial noise, and temporal jitter to tolerate imperfect Agent outputs. On LIBERO-Mem, without depth, online geometry, or planner-based object motion, 2AM reaches 76.3% average completion, a 61.5-point improvement over the strongest reported baseline of 14.8%, together with 63.0% relaxed and 11.8% strict success. These results show that task memory can remain Agent-side. They further show that Action Model capability depends not only on what the policy has learned, but on how precisely the Agent can steer it.

Related papers

• AgentAbstain: Do LLM Agents Know When Not to Act? [19.223843408018997]
  Existing evaluations mostly focus on task success rather than whether agents know when to abstain.<n>We present the first systematic evaluation framework for agentic abstention: the calibrated ability of tool-using LLM agents to recognize when not to act.<n>AgentAbstain is a paired-task benchmark built on an agent-native taxonomy of 8 abstention scenarios across pre-execution reasoning and runtime discovery.
  arXiv  Detail & Related papers  (2026-07-11T00:57:36Z)
• Agentic Abstention: Do Agents Know When to Stop Instead of Act? [13.780119058961331]
  LLM agents are expected to act over multiple turns, using search, browsing interfaces, and terminal tools to complete user goals.<n>We define Agentic Abstention, the problem of deciding when an agent should stop acting under uncertainty.<n>We study this problem across web shopping, terminal environments, and question answering, evaluating 13 LLM-as-agent systems and 2 agent scaffolds on more than 28,000 tasks.
  arXiv  Detail & Related papers  (2026-06-27T04:49:37Z)
• GTA: Generating Long-Horizon Tasks for Web Agents at Scale [82.43869456830664]
  We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control.<n>We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage.<n>Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.
  arXiv  Detail & Related papers  (2026-05-28T01:05:50Z)
• Workspace Optimization: How to Train Your Agent [49.579258311548635]
  We argue it is the agent's emphworkspace, the structured external substrate it reads, writes, and tests.<n>We propose a principled way to evolve the workspace, mirroring the structure of weight-space training.<n>We instantiate the idea in DreamTeam, a multi-agent harness for ARC-AGI-3.
  arXiv  Detail & Related papers  (2026-05-10T16:52:10Z)
• Beyond the Black Box: Interpretability of Agentic AI Tool Use [0.0]
  We introduce a mechanistic-interpretability toolkit built on Sparse Autoencoders and linear probes.<n>The framework reads model states before each action and infers both whether a tool is needed and how consequential the next tool action is likely to be.<n>We train the probes on multi-step trajectories from the NVIDIA Nemotron function-calling dataset and apply the same workflow to GPT-OSS 20B and Gemma 3 27B models.
  arXiv  Detail & Related papers  (2026-05-07T19:47:30Z)
• GenericAgent: A Token-Efficient Self-Evolving LLM Agent via Contextual Information Density Maximization (V1.0) [56.81743709880371]
  Long-horizon large language model (LLM) agents are fundamentally limited by context.<n>We argue that long-horizon performance is determined not by context length, but by how much decision-relevant information is maintained within a finite context budget.<n>We present GenericAgent (GA), a general-purpose, self-evolving LLM agent system built around a single principle: information density and context.
  arXiv  Detail & Related papers  (2026-04-18T17:59:15Z)
• ATP-Bench: Towards Agentic Tool Planning for MLLM Interleaved Generation [14.85015534787527]
  Interleaved text-and-image generation offers a more intuitive way to convey complex information.<n>Current paradigms rely on either image generation or retrieval augmentation, failing to unify factuality with creativity.<n>We argue that the next milestone in this field is Agentic Tool Planning, where the model serves as a central controller that autonomously determines when, where, and which tools to invoke to produce interleaved responses for visual-critical queries.
  arXiv  Detail & Related papers  (2026-03-31T15:47:59Z)
• AgentArk: Distilling Multi-Agent Intelligence into a Single LLM Agent [57.10083973844841]
  AgentArk is a novel framework to distill multi-agent dynamics into the weights of a single model.<n>We investigate three hierarchical distillation strategies across various models, tasks, scaling, and scenarios.<n>By shifting the burden of computation from inference to training, the distilled models preserve the efficiency of one agent while exhibiting strong reasoning and self-correction performance of multiple agents.
  arXiv  Detail & Related papers  (2026-02-03T19:18:28Z)
• ET-Agent: Incentivizing Effective Tool-Integrated Reasoning Agent via Behavior Calibration [68.89572566071575]
  ETAgent is a training framework for calibrating agent's tool-use behavior.<n>It is designed to progressively calibrate erroneous behavioral patterns to optimal behaviors.
  arXiv  Detail & Related papers  (2026-01-11T11:05:26Z)
• DeepAgent: A General Reasoning Agent with Scalable Toolsets [111.6384541877723]
  DeepAgent is an end-to-end deep reasoning agent that performs autonomous thinking, tool discovery, and action execution.<n>To address the challenges of long-horizon interactions, we introduce an autonomous memory folding mechanism that compresses past interactions into structured episodic, working, and tool memories.<n>We develop an end-to-end reinforcement learning strategy, namely ToolPO, that leverages LLM-simulated APIs and applies tool-call advantage attribution to assign fine-grained credit to the tool invocation tokens.
  arXiv  Detail & Related papers  (2025-10-24T16:24:01Z)
• ManiAgent: An Agentic Framework for General Robotic Manipulation [30.154478145473792]
  We introduce ManiAgent, an agentic architecture for general manipulation tasks.<n>Multiple agents involve inter-agent communication to perform environmental perception, sub-task decomposition and action generation.<n>ManiAgent achieves an 86.8% success rate on the SimplerEnv benchmark and 95.8% on real-world pick-and-place tasks.
  arXiv  Detail & Related papers  (2025-10-13T17:34:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.