产业动态 共 2815 条 · 筛选后 210 条

arXiv:2608.25757v2 Announce Type: replace Abstract: Generalist vision--language--action (VLA) policies learn long-horizon behavior mainly through short-horizon action prediction and reveal little beyond sampled commands. This creates two coupled bottlenecks: a single action target must implicitly absorb task progress, intermediate intent, and local reliability, while these control states remain hidden during execution. Inspired by functional principles of biological sensorimotor control, we intr…

arXiv:2510.25548v3 Announce Type: replace Abstract: In task and motion planning, high-level task planning is done over an abstraction of the world to enable efficient search in long-horizon robotics problems. However, the feasibility of these task-level plans relies on the downward refinability of the abstraction into continuous motion. When a domain's refinability is poor, task-level plans that appear valid may ultimately fail during motion planning, requiring replanning and resulting in slower…

arXiv:2608.26182v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used for verbal interaction in social robots, yet prompt design in human-robot interaction (HRI) remains underspecified. As a result, robots may present hallucinated capabilities, unclear behavioural boundaries, and misleading personas. This paper develops a framework for prompt design in LLM-based robots and introduces a structured prompt template comprising eight functional components through which …

arXiv:2608.27384v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models are increasingly promising for robotic manipulation, yet their real-world deployment remains bottlenecked by high inference latency and unstable asynchronous execution. This challenge is particularly pronounced in flow-matching-based VLA models, where action decoding requires multiple iterative steps conditioned on the VLM context. While efficient inference methods improve control frequency and asynchronous metho…

arXiv:2608.27079v1 Announce Type: new Abstract: Pretrained vision-language-action (VLA) policies provide strong priors for robot manipulation, yet adapting them online to fine-grained biomedical tasks remains challenging. Task success often hinges on subtle, view-dependent visual cues, while task-level rewards provide little guidance about which regions matter, making it difficult to learn task-relevant visual grounding from limited real-robot interaction. Online adaptation is further constraine…

arXiv:2608.26821v1 Announce Type: new Abstract: Vision-language-action (VLA) models leverage pretrained vision-language representations for robot control, yet simply adding historical frames does not reliably capture recent physical change. This is especially problematic in multi-stage manipulation, where visually similar states may require different actions depending on prior execution. To address this challenge, we present TemporalFlow-VLA, which learns compact execution history through physic…

arXiv:2608.26673v1 Announce Type: new Abstract: Large pretrained vision-language-action models dominate modern robot-manipulation benchmarks, but it remains unclear how much model scale is necessary for strong language-conditioned control, or whether fundamentally different control architectures can remain competitive at much smaller parameter budgets. We present PredVLA, a language-conditioned predictive-coding policy with only 0.68 million trainable network parameters and no robot-data pretrai…

arXiv:2608.26645v1 Announce Type: new Abstract: Vision-Language-Action Models~(VLAs) have demonstrated significant promise in generalizing to complex, long-horizon robotic manipulation tasks. However, their performance remains brittle, as they are typically trained on trajectory-monotonic, failure-free demonstrations. This reliance on ``perfect" data leaves them unable to recover from common execution errors, such as a missed grasp, a dropped object, or an unexpected collision. In this paper, we…

arXiv:2608.26578v1 Announce Type: new Abstract: This work introduces Configured Failure Trapping, a novel backdoor attack task against Vision-Language-Action (VLA) models, which aims to activate attacks through stealthy textual triggers and induce configured failure modes. Unlike prior backdoor attacks that treat any task failure as a successful attack, Configured Failure Trapping requires the attacker to control how the robot fails (e.g., causing the robot to grasp with a specified positional o…

arXiv:2608.26496v1 Announce Type: new Abstract: Navigation in unknown environments to find unforeseen objects has become increasingly feasible with capable vision and language foundation models. However, these models also introduce non-negligible inference latency, which becomes an important concern when agents must operate continuously in the real world. Most state-of-the-art methods are still developed in synchronous simulators, where the environment waits for the agent to act and inference ti…

站在 2026 年 8 月北京世界机器人大会（WRC）的展馆中心，如果你闭上眼，听到的不再是两年前那种对“机器人像人一样跳舞”的惊呼，取而代之的，是从业者们在心里拨动算盘，计算 ROI（投资回报率）的声音。 具身智能正在经历一场从“秀肌肉”到“翻账本”的集体变轨。 回看这半年的轨迹，这种趋势演进得极具层次感： 5 月的维也纳 ICRA，学术界和产业界第一次如此默契地把所有火力都集中在了 VLA（视觉-语言-动作）模型上，那是具身智能的“物理学补课期”； 7 月初的上海 WAIC，200 多家具身智能企业同台竞技，集体把演示从“后空翻”换成了“拧螺丝、叠衣服”，那是行业的“场景落地焦虑期”。 7 月底的悉尼 RSS，开源硬件、仿真工具链和遥操作数据方案成为主角，行业开始为“大脑聪明、身体廉价、数据贫瘠”搭建公共底座，那是产业的“基础设施奠基期”。 而在 8 月的北京 WRC，所有话题似乎都围绕工程约束和商业底线展开。在这里，没人关心你的模型是否有哲学意义上的“意识”，大家只盯着三个指标：多少钱？多稳？跑一个月能省几个人的工资？ “以前我们聊的是什么时候 AGI，现在我们在聊这台机器人跑一…

强化学习之父Rich Sutton接受红杉访谈，把矛头对准当下最主流的大模型路线：互联网数据终有上限、模型上线后权重基本冻结，行业或已陷入“局部最优”。他创办Oak Lab，押注能从自身经验持续学习的Agent，重新定义AI演进路径。 说起强化学习之父 Rich Sutton，AI圈几乎没人陌生。 他写下的《苦涩的教训》，至今仍被很多人视为理解AI发展的重要框架：从长期看，能够随着计算规模扩展的通用方法，往往会战胜依赖大量人工知识设计的方案。 但最近接受红杉资本访谈时，Sutton却把矛头对准了今天最主流的大模型路线。在他看来，LLM当然是一次伟大的科学突破，却也可能成为《苦涩的教训》的反面案例。 原因很简单。今天的大模型虽然摆脱了大量人工规则，却重新被“人类已经知道的东西”限制住了。互联网数据终究有限，合成数据背后依然需要人来决定什么值得生成。 而更关键的是，模型仍然不具备人类那样的持续学习能力。 一个司机开了10年车，会从经验里形成大量直觉。但今天的大模型，大部分学习都集中在预训练和后训练阶段。模型上线后，即使每天和几百万人交互，核心权重依然基本被冻结。 在Sutton看来，今天的…

从录音卡到儿童手表，再到摄像机，小度AI硬件掌舵人孙浩复盘了多款产品的成败得失。他提出，大模型能力涌现后，真正的挑战在于场景、成本、上下文与安全之间的平衡。本文从实战出发，拆解AI硬件从技术到产品成立的完整路径。 在 2026 AI 产品大会上，小度AI智能硬件掌舵人孙浩分享在 360 和小度做过的几条产品线，从复盘录音卡、儿童手表、摄像机和家庭 AI 伙伴的路径。他想回答的是能力涌现之后，怎样在场景、成本、上下文和安全之间把产品真正做成立。 以下内容根据现场音频和演示材料整理，Enjoy： 一、我做过很多硬件，但真正的问题一直没变 大家好，我是孙浩。 我一毕业就进了 360，在那里干了 12 年，最早做儿童手表，后来做智能音箱、边缘计算、视觉云，再后来整体负责 360 IoT 的产品研发、供应链和生产相关工作。年初我离开 360，到了小度。对我来说，这次分享不是站在旁观者角度讲一个趋势，而是把自己做过、踩过、犹豫过的产品拿出来复盘。 我给自己的定位一直是“极客型产品经理”。 这个词不是说我要做只有极客才会用的东西，而是说我希望自己有前沿技术的判断，把原来只在小圈子里流行、很晦涩的技术…

XPeng debuts the first major upgrade to its second-generation VLA, adding dynamic 4D spacetime understanding and bringing Robotaxi-grade L4 behavior to production cars.

Overnight undercutting by Zhipu's GLM-5.3-Flash and Alibaba's Qwen3.8-Flash is resetting what it means to be a flagship LLM in China's pricing war.

AI 硬件的较量，正从"接入 AI"转向"产品成立"——从 Demo 能跑，走向产品能卖、用户能留。 过去两年，大模型快速进入智能眼镜、机器人、AI 陪伴、AI 教育等终端。到了 2026 年，”接入 AI”已经不再新鲜，行业真正开始较量的，不再是 Demo 能不能跑起来，而是一个更传统、也更难的问题：这款产品为什么存在，能不能稳定量产，用户会不会长期使用。能力越多，产品越需要克制——多模态交互的真正难点，也从”能不能加上”，变成了”怎么取舍、怎么让用户觉得对”。 8月9日下午，声网联合人人都是产品经理，在北京渔阳饭店举办「AI 硬件多模态交互的产品化破局」专场沙龙，近百位产品人、开发者与创业者到场，共同拆解 AI 硬件从 Demo 走向真实产品的那道坎。 浙江湃启科技CEO孙思宁、声网AI产品线负责人姚光华、灵宇宙产品负责人刘翠涛、阿里云百炼应用高级产品经理司红江，以及移远通信北区技术总监武风停，从产品、交互、端侧与供应链等不同位置，围绕 AI 硬件从 Demo 走向真实产品展开讨论。以下为几位嘉宾分享的核心内容。 一、具身在神不在形：先定义产品，再谈模型 孙思宁老师作为语音AI领域…