智驾算法

Search documents
千里智驾的软硬一体
自动驾驶之心· 2025-10-17 00:03
以下文章来源于红色星际 ,作者红色星际科技 据悉,千里智驾或将与爱芯元智在智驾芯片上进行合作,爱芯元智的芯片匹配千里智驾的算法。 作为顶流牌面的千里智驾,在完成了对极氪智驾团队和迈驰智行团队合并之后,在算法上的布局初步成型。 而且千里智驾也发布了宏大的技术目标, 覆盖L2+、L3和Robotaxi。 红色星际 . 让更多人,更深入地了解自动驾驶行业! 作者 | 红色星际科技 来源 | 红色星际 点击下方 卡片 ,关注" 自动驾驶之心 "公众号 戳我-> 领取 自动驾驶近30个 方向 学习 路线 >>自动驾驶前沿信息获取 → 自动驾驶之心知识星球 本文只做学术分享,如有侵权,联系删文 自动 驾 驶 之心 -END- 自动驾驶 毕业论文 不会选方向 无人辅导( 自动驾驶2 键上岸! 不过,千里智驾的宏大叙事中还缺芯片这一环节,也就是软硬一体。 华为是软硬一体交付,Momenta自研芯片也有一定进展,地平线也发布了基于J6P 的城区方案HSD。 如果在芯片上不采取自研的方式,寻找一家能够紧密配合算法的芯片公司就很有必要。 另外,面向下一波软硬技术大迭代也需要有能够紧密合作的芯 片公司。 无论特斯拉还是国内的新势力以 ...
【重磅深度】AI+汽车智能化系列之十一——以地平线为例,探究第三方智驾供应商核心竞争力
东吴汽车黄细里团队· 2025-05-09 12:01
Core Viewpoint - The company is optimistic about the breakthrough opportunities for leading third-party intelligent driving suppliers, driven by the demand for equal access to intelligent driving and the performance catch-up and mass production validation [2][8]. Group 1: Market Opportunities - Leading third-party intelligent driving suppliers are expected to become the optimal solution for second- and third-tier automakers seeking equal access to intelligent driving, with a potential market share of around 50% of total new car sales [2][8]. - The current trend of intelligent driving is accelerating towards equal access, with a focus on cost reduction in systems, as automakers balance performance and cost in their strategies [2][8]. Group 2: Domestic Chip Comparison - NVIDIA's Orin series chips currently dominate the high-end intelligent driving market, but domestic chip suppliers have made significant progress in performance, mass production validation, and customer acquisition over the past five years [3][39]. - The domestic chip leader, Horizon Robotics, is entering a new cycle of product iteration and business model elevation, leading in the rollout of mid- to high-end intelligent driving chips and algorithms [11][39]. Group 3: Core Value of Third-Party Chip Suppliers - The importance of being a first mover is highlighted, as intelligent driving chips typically require over three years of R&D and manufacturing cycles, necessitating continuous iteration capabilities for cost-performance balance [4][54]. - The design and manufacturing cost perspective indicates that a 7nm intelligent driving chip can achieve cost parity with mature chip procurement at a production volume of 1.5 million units, emphasizing the need for high output and rapid iteration capabilities for self-developed chips [4][57]. Group 4: Algorithm Insights - The "BEV + Transformer" algorithm approach, focusing on "heavy perception, light mapping," has been validated and widely applied, reducing risks for Tier 1 suppliers and allowing them to keep pace with cutting-edge technologies [4][62]. - Horizon Robotics' latest intelligent driving algorithm, HSD, is positioned as a "showcase," balancing performance and efficiency while addressing the challenges of scaling up and out in intelligent driving systems [62][63]. Group 5: Industry Trends - The intelligent driving landscape is expected to see a significant shift towards equal access by 2026, with many domestic automakers planning to adopt domestic chips as their mainstream solution [28][43]. - The competitive landscape among automakers is intensifying, with a focus on high-level intelligent driving capabilities and the need for cost-effective solutions [2][8].