{
 "date": "2026-07-07",
 "count": 3,
 "signals": [
  {
   "key": "1be2ff8d9601acae",
   "date": "2026-07-07",
   "dimension": "research",
   "score": 2.8,
   "rules_hit": [
    "R7-research",
    "cross-source×2"
   ],
   "title_en": "Embodied.cpp: Portable Inference Runtime for Vision-Language-Action Models on Edge Devices",
   "title_zh": "Embodied.cpp：异构边缘设备上的嵌入式AI推理运行时",
   "summary_en": "1. Researchers presented Embodied.cpp, a C++ inference runtime designed to enable efficient deployment of vision-language-action (VLA) and world-action models (WAMs) on heterogeneous edge devices and robots.\n2. The runtime organizes execution into five modular layers (input adapters, sequence builders, backbone execution, head plugins, deployment adapters) to solve fragmentation across model-specific Python stacks, supporting multi-rate execution and latency-optimized batch-1 inference.\n3. The infrastructure addresses closed-loop control requirements with real-time responsiveness for embodied AI deployment, a pattern critical to autonomous vehicle perception-action loops.",
   "summary_zh": "1. 研究人员提出了Embodied.cpp，这是一个C++推理运行时，用于支持视觉-语言-动作（VLA）和世界-动作模型（WAM）在异构边缘设备和机器人上的高效部署。\n2. 该运行时将执行过程组织为五个模块化层（输入适配器、序列构建器、骨干执行、头部插件、部署适配器），以解决现有模型特定Python栈的碎片化问题，支持多速率执行和延迟优先的批量1推理。\n3. 该基础设施解决了嵌入式AI模型对闭环控制和实时响应性的需求，这是自动驾驶感知-动作循环的关键需求。",
   "expert_take_en": "",
   "expert_take_zh": "",
   "expert_name": "",
   "sources": [
    {
     "source_id": "arxiv-ad",
     "tier": "T1",
     "title": "Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots",
     "url": "https://arxiv.org/abs/2607.02501v2",
     "published": "2026-07-02T17:58:28Z"
    },
    {
     "source_id": "arxiv-ad",
     "tier": "T1",
     "title": "Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots",
     "url": "https://arxiv.org/abs/2607.02501v1",
     "published": "2026-07-02T17:58:28Z"
    },
    {
     "source_id": "hf-daily-papers",
     "tier": "T1",
     "title": "Embodied.cpp: A Portable Inference Runtime of Embodied AI Models on Heterogeneous Robots",
     "url": "https://huggingface.co/papers/2607.02501",
     "published": "2026-07-01T20:00:00"
    }
   ]
  },
  {
   "key": "eb4cb85e1f70f37b",
   "date": "2026-07-07",
   "dimension": "research",
   "score": 2.8,
   "rules_hit": [
    "R7-research",
    "cross-source×2"
   ],
   "title_en": "ACID Framework Improves World Model Planning via Inverse Dynamics Action Consistency",
   "title_zh": "ACID框架通过逆向动力学验证行动一致性，改进世界模型规划",
   "summary_en": "1. ACID is a decision-time planning framework for embodied control that uses action-conditioned world models to search over candidate action sequences.\n2. It adds an action consistency cost via inverse dynamics that verifies predicted trajectories are actually realizable in the environment, not just goal-optimized.\n3. Standard planning judges candidates only by terminal state proximity to goal; ACID additionally checks whether intermediate transitions are executable by inferring actions that would explain each predicted transition.",
   "summary_zh": "1. ACID是用于具身控制的决策时规划框架，采用行动条件化世界模型搜索候选动作序列。\n2. 它通过逆向动力学添加行动一致性成本，以验证预测轨迹在环境中的实际可实现性，而非仅优化目标。\n3. 标准规划仅根据终端状态与目标的接近度评判，ACID额外检查中间转移的可执行性，通过推断能解释每个预测转移的动作实现。",
   "expert_take_en": "",
   "expert_take_zh": "",
   "expert_name": "",
   "sources": [
    {
     "source_id": "hf-daily-papers",
     "tier": "T1",
     "title": "ACID: Action Consistency via Inverse Dynamics for Planning with World Models",
     "url": "https://huggingface.co/papers/2607.02403",
     "published": "2026-07-01T20:00:00"
    },
    {
     "source_id": "arxiv-ad",
     "tier": "T1",
     "title": "ACID: Action Consistency via Inverse Dynamics for Planning with World Models",
     "url": "https://arxiv.org/abs/2607.02403v1",
     "published": "2026-07-02T16:38:10Z"
    }
   ]
  },
  {
   "key": "f5e534aad44b0466",
   "date": "2026-07-07",
   "dimension": "research",
   "score": 2.8,
   "rules_hit": [
    "R7-research",
    "cross-source×2"
   ],
   "title_en": "GigaWorld-1: World Models Benchmark for Embodied Policy Evaluation",
   "title_zh": "GigaWorld-1：用世界模型评估具身AI策略的基准",
   "summary_en": "1. Researchers introduce GigaWorld-1, a systematic study and benchmark (WMBench) using world models to evaluate embodied robot policies, eliminating the need for expensive real-world rollouts.\n2. WMBench uses real-robot teleoperation data across diverse manipulation tasks, analyzing 7 video world models and 4 action representation schemes to understand which properties make world models reliable for policy assessment.\n3. World models address a critical bottleneck in evaluating embodied foundation models—unlike LLMs assessed via digital benchmarks, robot policies traditionally require slow, costly real-world testing constrained by hardware and human supervision.",
   "summary_zh": "1. 研究团队推出GigaWorld-1，这是一项系统性研究和基准测试（WMBench），使用世界模型评估具身机器人策略，消除对昂贵真实世界测试的需求。\n2. WMBench使用真实机器人遥操作数据，涵盖多样化操纵任务，分析7个视频世界模型和4种动作表示方案，以理解哪些特性使世界模型在策略评估中可靠。\n3. 世界模型解决了评估具身基础模型的关键瓶颈——与通过数字基准测试评估大语言模型不同，机器人策略传统上需要缓慢、成本高的真实世界测试，受硬件和人类监督限制。",
   "expert_take_en": "",
   "expert_take_zh": "",
   "expert_name": "",
   "sources": [
    {
     "source_id": "arxiv-ad",
     "tier": "T1",
     "title": "GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation",
     "url": "https://arxiv.org/abs/2607.02642v1",
     "published": "2026-07-02T17:02:43Z"
    },
    {
     "source_id": "hf-daily-papers",
     "tier": "T1",
     "title": "GigaWorld-1: A Roadmap to Build World Models for Robot Policy Evaluation",
     "url": "https://huggingface.co/papers/2607.02642",
     "published": "2026-07-01T20:00:00"
    }
   ]
  }
 ]
}