Core Viewpoint - DrivePI is introduced as a novel unified spatial-aware 4D multimodal large language model (MLLM) framework that integrates coarse-grained language understanding with fine-grained 3D perception capabilities, bridging the gap between vision-based and VLA paradigms in autonomous driving [2][38]. Group 1: Project Overview - DrivePI is developed collaboratively by Hong Kong University, leading the project with contributions from companies like Huawei and universities such as Tianjin University and Huazhong University of Science and Technology [2]. - The model is designed to perform spatial understanding, 3D perception, prediction, and planning tasks through end-to-end optimization, showcasing its capability to handle complex autonomous driving scenarios [4][6]. Group 2: Technical Innovations - DrivePI incorporates a multimodal perception approach, utilizing LiDAR alongside camera images to enhance spatial understanding and provide accurate 3D geometric information [11]. - The model generates intermediate fine 3D perception and prediction representations, ensuring reliable spatial awareness and enhancing the interpretability and safety of autonomous driving systems [11]. - A rich data engine is developed to seamlessly integrate 3D occupancy and flow representations into natural language scene descriptions, allowing the model to understand complex spatiotemporal dynamics [11]. Group 3: Performance Metrics - DrivePI outperforms existing VLA models, achieving a 2.5% higher average accuracy on nuScenes-QA compared to OpenDriveVLA-7B and reducing collision rates by 70% from 0.37% to 0.11% [5][16]. - In 3D occupancy and flow prediction, DrivePI achieved 49.3% OccScore and 49.3% RayIoU, surpassing the FB-OCC method by 10.3 percentage points [15][21]. - The model demonstrated a 32% reduction in L2 error for trajectory planning compared to VAD, showcasing its effectiveness in planning tasks [16]. Group 4: Data Engine and Annotation - The data engine for DrivePI operates in three main stages, focusing on generating diverse question-answer pairs for 4D spatial understanding and planning reasoning [12][18]. - Scene understanding annotations are generated to avoid confusion in distinguishing different views, enhancing the model's ability to interpret various perspectives [18]. Group 5: Ablation Studies and Insights - Ablation studies indicate that combining text and visual heads improves performance across most tasks, demonstrating the effectiveness of unifying text understanding with 3D perception, prediction, and planning [23]. - The impact of different text data scales was explored, revealing significant improvements in occupancy state prediction accuracy when increasing the training data size [26]. Group 6: Future Prospects - DrivePI is expected to inspire future research directions in autonomous driving by enhancing the interpretability and decision-making capabilities of systems through language reasoning and detailed 3D outputs [38].

一次大规模停电，暴露了全球无人车一哥的短板。 被曝估值冲上千亿美元没几天，Waymo就因为当地停电全面停摆了，挡在路中间，造成城市拥堵，相关视频疯传。 马斯克第一时间"补刀"，表示自家Robotaxi就没受到影响。看上去，特斯拉代表的L2渐进式路线，似乎小胜了一局……反正马哥认为这就是彰显优越性的 时刻。 在Robotaxi战场上，今年马斯克的一举一动，都把自动驾驶之争推向了新的高潮，大洋两岸更多玩家开始入场，沿着「特斯拉路线」前进，与「Waymo路 线」争夺自动驾驶圣杯。 所以问题是，停电是如何影响Waymo Robotaxi的？ 当地停电，Waymo停工 Waymo停摆源自一场火灾，旧金山变电站失火，导致当地大规模停电，据说直接影响到13万居民用电。 更要命的是，因为大范围停电，马路上的红绿灯都不亮了，引发Waymo无人车全面停摆。 真是屋漏偏逢连夜雨，本就混乱的交通，这下因为无人车挡在路上变得更堵了。Waymo只好连夜找拖车运走了无人车，同时宣布在当地停运，目前还不 清楚什么时候重新上线。 所以为啥停电会导致Waymo停运？首先是官方回应暴露的运营漏洞。 Waymo在官方声明中解释称，Waymo的Rob ...

Core Viewpoint - Multiple companies have publicly questioned Waymo's technological approach to Robotaxi following a significant incident in San Francisco where a power outage caused traffic disruptions and highlighted the limitations of Waymo's system [1][3][4]. Group 1: Incident Overview - On December 20, a fire at a PG&E substation led to a power outage affecting over 30% of San Francisco, impacting more than 130,000 residents [1]. - The outage caused traffic lights to go dark, resulting in Waymo's Robotaxis becoming immobilized in the streets, contributing to traffic chaos [1][3]. Group 2: Industry Reactions - Executives from companies like Tesla, Horizon, and Xpeng have raised concerns about Waymo's reliance on high-definition maps and pre-programmed rules, suggesting that this approach is inadequate for handling unexpected situations [3][6][8]. - Elon Musk stated that Tesla's Robotaxi was unaffected during the power outage, contrasting it with Waymo's performance [4][6]. Group 3: Technological Critique - Waymo's L4 Robotaxi technology is criticized for its rigid response to unforeseen events, as evidenced by a previous incident where it failed to stop for a school bus [6]. - In contrast, Tesla and other companies advocate for AI-driven models that learn from vast amounts of data, aiming for a more adaptable and robust autonomous driving solution [6][8]. Group 4: Future Directions - Horizon's CEO emphasized the need for AI advancements to achieve true safety in autonomous driving, arguing that relying on rules and infrastructure is insufficient [8][10]. - Xpeng's senior director highlighted the importance of developing vehicles that can understand and react to real-world scenarios like a human driver, indicating a shared vision with Tesla regarding the future of Robotaxi technology [10][13].

编辑：冷猫 上周六，美国旧金山发生了一次大规模的停电事故。 这次停电事件似乎是由该市一处 太平洋煤气与电力公司（Pacific Gas & Electric，PG&E）变电站发生火灾所引发的。SFGate 报道称，约有 12 万名用户受到此次停 电影响。 不过，我们关注的并不只是停电事故，而是停电导致的一个非常诡异的现象： Waymo 自动驾驶汽车在停电期间大规模瘫痪。 Waymo 在周六表示，由于停电，该市暂时停止了服务。直到周日下午晚些时候，Waymo 发言人表示公司已恢复服务。 在此期间，有大量照片和视频被发布到社交媒体上，拍下了 Waymo 自动驾驶出租车在道路中间和十字路口停车。也说明了这一瘫痪现象并非个例，而是普遍现 象。 人类司机要么被堵在它们后面，要么绕过它们，造成了严重的交通堵塞现象。 Nic Cruz Patane @niccruzpatane Subscribe Q This is insane, Waymo vehicles all over SF are bricked due to a power outage. 这太疯狂了，由于停电，旧金山的所有 Waymo 车辆都变砖了。 0: ...

12月21日，全球Robotaxi龙头企业Waymo因在多个十字路口大面积停车而备受关注。 此次事故的起因是旧金山当地的红绿灯受到停电的影响而失灵。 Waymo的Robotaxi由于无法灵活应对熄灭的红绿灯这类突发事故，而做出了在十字路口停车这一看似 更"安全"的操作。 此次事故也给国内将要狂飙的Robotaxi车队带来更多警示意味。 回溯整个2025年，中国Robotaxi赛道的资本化被按下了加速键。 据信风不完全统计，今年以来已有哈啰、滴滴、小马智行、文远知行等多家公司完成新一轮融资，融资 额合计已超百亿。 这为各家2026年的Robotaxi车队扩充补充了充足的弹药。 信风结合当前已披露2026年目标车队规模的企业测算，2026年仅小马智行、文远知行和滴滴的Robotaxi 车队规模合计便有望达到8000辆。 中国Robotaxi正在逼近万辆这一里程碑节点。 但与竞对形成反差的是，作为国内Robotaxi NO.1的百度萝卜快跑却是异常的低调。 迄今为止，萝卜快跑仍只是百度的一个业务部门，并未独立进行对外融资，这也给外界留下了无限的遐 想。 这种"克制"背后，萝卜快跑同样在筹划扩大运营规模。近期李彦宏在 ...

Investment Rating - The report maintains a positive outlook on the computer industry, recommending key targets such as Hand Enterprise, Wuxi Unicomp Technology Co., Ltd., Beijing Kingsoft Office Software, Newland Digital Technology, Hikvision, Hygon, and Dawning Information Industry [3][9]. Core Insights - The AI foundation model industry in China is experiencing accelerated productization and capitalization, with significant advancements from Volcengine, which released Doubao 1.8 and Seedance 1.5 Pro, achieving global leadership in multimodal capabilities and processing over 50 trillion tokens daily [10]. - Zhiyuan AI and MiniMax have successfully passed HKEX hearings and are expected to list in early 2026, providing a quantifiable market valuation benchmark for the industry [10]. - The Ministry of Industry and Information Technology has issued L3 autonomous driving permits, marking a transition to conditional commercialization, with specific models from Changan and BAIC Arcfox approved for limited road use [11]. - Domestic GPU companies are actively pursuing capital market entry to support R&D and market expansion, with TianShu ZhiXin planning to raise 300 to 400 million USD and other companies like Biren Technology and Kunlun Chip Technology advancing their listings [12]. Summary by Sections AI Foundation Models - The report highlights the rapid advancements in AI technology, particularly with the release of Doubao 1.8 and Seedance 1.5 Pro, which have significantly reduced operational costs for enterprises [10]. - The successful IPO preparations of Zhiyuan AI and MiniMax are expected to enhance the industry's financial landscape and provide a basis for future valuations [10]. Autonomous Driving - The issuance of L3 autonomous driving permits signifies a critical step towards commercial applications, with safety regulations emphasizing the responsibility of manufacturers over drivers [11]. GPU Market Developments - The report notes a surge in domestic GPU companies seeking public listings to fund advanced chip development, reflecting strong market interest and a shift towards self-reliance in semiconductor technology [12].

↑点击上面图片 加雪球核心交流群 ↑ 今天A股高开高走！截至收盘，沪指涨0.69%，深成指涨1.47%，创业板指涨2.23%。 沪深两市成交额1.86万亿，较上一个交易日放量1360亿。盘面上，市场热点快速轮动，全市场超2900只个股上涨，其中有105只个股涨停。海南 自贸区、贵金属、算力硬件等板块涨幅居前，医药商业、影视院线、教育、银行等板块跌幅居前。一起来看今天热点。 01 贵金属狂飙， 黄金、白银齐创新高！ 在降息预期和地缘政治紧张局势的双重推动下，贵金属集体强势， 现货黄金继续攀升，站上4400美元，刷新历史新高水平，伦敦白银现货站上69 美元，年内暴涨近140%，创下历史新高， 广期所钯期货主力合约涨停，日内涨7.00%，再创上市以来新高，为上市以来第三次涨停。 A股贵金属板块活跃，湖南白银、晓程科技涨超7%，西部黄金、中金黄金、山东黄金、四川黄金、赤峰黄金、招金黄金、湖南黄金等跟涨。 | | | 贵金属 | | | | --- | --- | --- | --- | --- | | | | 2944.78 (+118.25 +4.18%) | | | | 讨论 | 板块分析 | 成分股 | 相关E ...

优步于当地时间周一宣布，2026 年上半年将在伦敦启动一项使用百度 Apollo Go RT6 自动驾驶出租车的 试点项目，预计明年年底前可在该市正式投入服务。 【#百度Uber在英国测试无人驾驶网约车#，明年年底前提供服务】Uber表示，将与百度公司合作，在英 国开展无人驾驶网约车试点项目。此前，美国自动驾驶技术公司 Waymo 已于本月在伦敦启动了相关测 试。 ...

人工智能AIETF（515070）跟踪CS人工智能主题指数（930713），成分股选取为人工智能提供技术、 基础资源以及应用端个股，聚集人工智能产业链上中游，俗称"机器人"大脑"缔造者"，万物互联"地 基"。前十大权重股包括中际旭创、新易盛、寒武纪-U、中科曙光、科大讯飞、豪威集团、海康威视、 澜起科技、金山办公、紫光股份等国内科技龙头。 A股科技赛道午后延续强势，CPO、无人驾驶、人工智能、人形机器人、存储板块全线上涨，沪市规模 最大的人工智能AIETF（515070）盘中涨2.08%，其持仓股新易盛、中际旭创涨幅均超6%，北京君正、 浪潮信息、芯原股份等股涨幅居前。 消息方面，据南方财经报道，深圳功能型无人车投放总量已达 1218 台（其中包含无人物流车 825 台），车辆数在全国一线城市中位列第一，累计开放运行线路 1594 条、总里程达 5534 公里。今年以 来，深圳等地颁发 L4 级自动驾驶微型货车测试牌照，标志着自动驾驶货运服务进入多场景、规模化发 展阶段。报道还提到，我国拥有全球最大的自动配送市场。截至 2025 年上半年，全国已有 103 个城市 开放无人配送车辆路权，全国无人配送车保有量突 ...

Market Overview - On December 22, 2025, the market closed higher with a trading volume of 1.75 trillion, an increase of approximately 70 billion from the previous trading day[1] - The majority of industries saw gains, particularly commerce, light industry, environmental protection, and social services, while a few sectors like banking, coal, and electronics experienced slight declines[1] - The Shanghai Composite Index showed a three-day upward trend, indicating some technical recovery, although the Sci-Tech Innovation Board remains in a weaker technical state[1] Industry Trends - Recently active sectors such as cyclical and consumer industries have shown increased activity, contrasting with the historically dominant technology sector[1] - The commercial aerospace sector remains strong but has shown signs of slowing momentum, with notable stock differentiation within the sector[1] - Valuations in the non-ferrous metals and insurance sectors are worth further attention, while the autonomous driving sector has seen a recent accumulation of events[1] Fund Flow - On December 19, the net inflow for the Shanghai Stock Exchange was 24.489 billion, while the Shenzhen Stock Exchange saw a net outflow of 8.336 billion[4] - The top three sectors for net inflow were automotive parts, general equipment, and general retail, while the semiconductor and electronic components sectors faced the largest outflows[4] Economic Indicators - From January to November 2025, the national online retail sales increased by 9.1%, with smart wearables and smart robots growing by 22.1% and 19.4%, respectively[11] - The national railway transported 3.727 billion tons of goods from January to November, reflecting a year-on-year growth of 2.7%[12] Investment Activity - In December, the stock private equity positions reached a new high for the year, with the stock private equity position index rising to 83.59%, marking a 0.61 percentage point increase from the previous week[13] - Public fund institutions participated in 85 individual stock placements this year, with a total allocation exceeding 34.088 billion, a 14.24% increase from the same period last year[15]