ScholarPulse 日报 2026-06-23

2026-06-23 学术简报：2 篇。HoloAgent-0是一个统一的具身智能体框架，通过3D空间记忆实现真实机器人部署。

今日速览

序号	标题	来源	日期	主题	推荐等级
1	HoloAgent-0: A Unified Embodied Agent Framework with 3D Spatial Memory	arXiv	2026-06-22	AI-Agent	高
2	Self-Compacting Language Model Agents	arXiv	2026-06-22	AI-Agent	高

重点论文与技术动态

1. HoloAgent-0: A Unified Embodied Agent Framework with 3D Spatial Memory

来源：arXiv
日期：2026-06-22
作者/机构：Xiaolin Zhou, Liu Liu, Tingyang Xiao, Wei Feng, Fa Fu, Xinrui Meng
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.RO, cs.CV

一句话结论

HoloAgent-0是一个统一的具身智能体框架，通过3D空间记忆实现真实机器人部署。

核心内容

Embodied AgentOS将语言指令转换为可执行技能图，动态调度机器人资源，实时监控执行并触发反馈澄清或重规划。
采用三层耦合架构组织异构机器人：Embodied AgentOS层（闭环执行）、3D空间记忆层（物理世界锚定）、具身技能层（机器人动作）。
在真实硬件部署，验证了空间记忆、长周期导航及闭环执行在运动生成、物体搜索、跨机器人协调和移动操作任务中的有效性。

方法与数据

方法基于Embodied AgentOS和3D空间记忆技术构建框架。
摘要未明确具体实验数据细节。

价值判断

值得关注：统一解决物理机器人执行的连续性、依赖性、不确定性及安全约束问题。
可复用点：三层架构设计可迁移至其他具身AI系统，提升模块集成度。
局限/待核查：摘要未明确框架局限性或待验证点。

摘要

LLM agents follow a practical execution loop in digital environments: they reason over structured states, invoke tools, inspect feedback, and revise actions. Extending this loop to physical robots is difficult because physical execution is continuous, embodiment-dependent, uncertain, and constrained by safety. Existing embodied-AI systems have advanced manipulation, spatial understanding, navigation, and humanoid control, but these capabilities often remain specialized modules or loosely coupled decision loops. In this work, we introduce HoloAgent-0, a unified embodied agent framework for real-world robot deployment. Embodied AgentOS converts language instructions into executable skill graphs, schedules robot resources, monitors execution, and triggers clarification or re-planning from runtime feedback. HoloAgent-0 organizes heterogeneous robot models and controllers through three coupled layers: Embodied AgentOS for closed-loop execution, 3D spatial memory for physical world grounding, and embodied skills for robot action. We deploy HoloAgent-0 on real hardware and evaluate its spatial memory, long-horizon navigation, and closed-loop execution across motion generation, object search, cross-robot coordination, and mobile manipulation.

2. Self-Compacting Language Model Agents

来源：arXiv
日期：2026-06-22
作者/机构：Tianjian Li, Jingyu Zhang, William Jurayj, Xi Wang, Chuanyang Jin, Mehrdad Farajtabar
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.CL

一句话结论

SelfCompact通过模型自决策实现上下文自适应压缩，显著降低token成本并提升任务性能。

核心内容

长代理轨迹积累过时内容，锚定后续生成并超出上下文窗口，导致性能下降。
SelfCompact结合压缩工具（模型调用总结上下文）和轻量级规则（定义触发时机如子任务解决或轨迹收敛），实现无需微调的自适应压缩。
实验在6个基准测试上，数学性能提升18.1点，代理搜索提升5-9点，token成本降低30-70%，并暴露元认知差距（模型无法自判断上下文状态）。

方法与数据

方法：模型在推理时调用压缩工具总结上下文，并基于轻量级规则动态决定压缩时机。
数据：6个基准测试（竞争性数学和代理搜索）及7个模型。

价值判断

值得关注：无需微调解决上下文膨胀问题，揭示模型元认知缺陷并提供轻量级解决方案。
可复用点：轻量级规则可直接应用于其他模型，无需额外训练或外部监督。
局限/待核查：摘要未明确具体局限。

摘要

Long agent traces composed of chains of thought and tool calls accumulate stale content that anchor subsequent generations, and eventually outgrow the context window. Existing scaffolds mitigate it with fixed-interval compaction triggered at a token threshold. Such triggers pay no heed to trajectory structure, risking discard of partial results mid-derivation or mid-search. We propose SelfCompact, a scaffold that allows the model itself to decide when and how to compact. Specifically, it pairs two inference-time elements: (i) a compaction tool the model invokes to summarize the accumulated context, and (ii) a lightweight rubric specifying when to fire (a sub-task has resolved, or the trajectory is converging) and when to suppress (mid-derivation, or when stuck). Both are needed. The tool alone is unevenly used across open-weight models, often invoked at unhelpful moments or not at all; the rubric alone cannot act. Together, they elicit effective adaptive compaction without any fine-tuning or external supervision. We present empirical results on six benchmarks (competitive math and agentic search) and seven models. Our results show that SelfCompact matches or exceeds fixed-interval summarization at a fraction of the token cost, improving over a no-summarization baseline by up to 18.1 points on math and 5-9 points on agentic search at 30-70% lower per-question cost. Our results expose a meta-cognitive gap: although unprompted models cannot reliably tell when their own context is rotting, a lightweight rubric closes this gap, reframing when to compact as a capability that scaffolds can supply without training.