达摩院王刚首次对外解读阿里自动驾驶布局，阿里无人车配送快递已突破100万单

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"自动驾驶技术的发展不能再靠讲故事、画大饼和资本烧钱的方式来推进，业界需要看到清晰的商业模式"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"阿里自动驾驶实现规模化落地，无人车配送快递突破100万单"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"9月27日，阿里巴巴集团副总裁、达摩院自动驾驶实验室负责人王刚宣布，达摩院自研的L4级自动驾驶产品、末端物流无人车“小蛮驴”已落地全国22个省份，累计配送订单超100万单，为20多万人送过快递。3年后，小蛮驴车队规模将达1万辆，有望日均配送包裹100万件。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/f2\/ae\/f269ecf82870cb85367f6877caea4eae.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"小蛮驴是阿里为“最后三公里”研发的自动驾驶产品，服务于快递、外卖、生鲜及各类即时配送需求。2020年9月问世后，达摩院与菜鸟合作，率先针对快递配送形成服务闭环，将高校、社区、园区的快递按用户预约时间送货到楼。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/2d\/5f\/2d35969ba983fea2fa7fa9f9b262ce5f.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"“小蛮驴”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"这是王刚首次对外解析阿里巴巴自动驾驶战略布局。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2017年达摩院成立，自动驾驶实验室承担起阿里集团自动驾驶技术的整合统一任务，由王刚带队组建研发力量，开始面向应用的技术探索。王刚强调，阿里将持续长期投入自动驾驶。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"自动驾驶技术发展了三十多年，终于开始大规模应用，进入普通人的日常生活。王刚说：“自动驾驶已经进入产品驱动的发展阶段。小蛮驴的100万单只是一个起点，它为自动驾驶技术的进一步发展和落地打下了很好的基础。”"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"100万订单背后"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在接受InfoQ等媒体采访时，王刚表示，小蛮驴快递订单突破100万意味着自动驾驶的规模化应用实现了从0到1的突破，在产能爬坡最艰难的早期阶段，达摩院自动驾驶实验室逐步克服了技术、成本、供应链等多方面的挑战，验证末端物流这一垂直赛道。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"技术挑战"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"无人车技术涵盖算法、算力、硬件、系统等多个维度，算法能力最能体现无人车的智能水平。算法研发依赖数据驱动，需要处理大规模的场景数据、运营数据。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"达摩院自动驾驶实验室通过强化学习、智能仿真等手段，掌握了一套对数据进行结构化分析和处理的能力，比如，提出了“小前台、大中台”理念，首创自动驾驶“算法中台”AutoDrive，让计算机自动学习并找到适合每种算法模型的结构和参数，用更符合计算机思维规律而非人类思维规律的方式处理数据，“以计算换智能”。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"场景挑战"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"末端物流场景远比人们想象的复杂。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不同于规则明确、参与者有限的公开道路“结构化场景”，末端环境是典型的“非结构化场景”，交通参与者包括人、小动物、非机动车、机动车、各种路障和临时障碍物等，没有规范的车道设计和通行规则，无人车与任何道路参与者的交互都是复杂的博弈问题，比如窄路会车，无人车需要辨别对方意图并作出“你进我退”、“你退我进”等的行车决策。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"成本挑战"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"以大规模应用为目标的自动驾驶产品，如果不能将成本降低到量产水平，一切都是纸上谈兵。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从末端场景的社会需求来说，“成本可控”也是无人车的应用前提之一。我们通过核心算法自研+核心硬件深度定制的方式，大幅降低了无人车的研发制造成本。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"小蛮驴的核心传感器、底盘和嵌入式计算平台全部采用深度定制，达摩院高性能、低功耗、低成本的嵌入式异构计算单元，能以1\/3算力的达到同等智能水平。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，达摩院自动驾驶实验室还逐步克服了无人车量产和规模化运营过程中遇到的供应链、运维等多种挑战，在从0到1的阶段，逐步推动产能爬坡。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"直接定位L4级路线，“直奔”商业验证"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"自动驾驶技术百家争鸣，总体来说分为两大路线：路线一：从L2到L4的纵向技术迭代路径，比如特斯拉。从辅助驾驶技术起步，固定场景，向上逐步实现技术迭代。路线二：从垂直场景L4到泛化场景L4，横向场景拓展路径。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不同于从L2到L4的纵向技术迭代路线，阿里自动驾驶走的是L4级横向场景拓展路线，自始至终定位L4级技术，从垂直领域起步，在非载人驾驶场景下完成技术累积，逐步切换到无人卡车、乘用车、移动机器人等泛化场景。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“自动驾驶技术的发展不能再靠讲故事、画大饼和资本烧钱的方式来推进，业界需要看到清晰的商业模式”，王刚在接受媒体采访时表示，“因此，阿里选择了一条“技术高打”、“场景下沉”的路线，希望少走弯路，且快速完成末端物流赛道的商业验证”。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"阿里无人卡车“大蛮驴”已启动研发，与小蛮驴共享同一技术框架"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"王刚表示，面向公开道路的无人卡车“大蛮驴”也已启动研发。“大蛮驴”是一款适用于城配物流的轻型卡车，应用场景包括城市快递网点-快递末端节点之间的物流配送等。目前，达摩院和菜鸟合作，已展开无人轻卡路测。阿里计划在3年内让无人卡车出现在上万条城市配送路线上，开展常态化运营。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比起物流车，无人卡车的研发面临着更大的技术挑战，但小蛮驴已经沉淀的技术能力和场景数据为无人轻卡的快速研发奠定了基础。大小蛮驴共享同一技术框架，达摩院自研的自动驾驶机器学习平台AutoDrive、云上智能仿真测试平台、嵌入式计算单元以及深度定制传感器技术，都已顺滑“平移”到了无人卡车的技术栈。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/bb\/76\/bbdb59786be2449df0f7b9f1be891576.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"italic"}],"text":"小蛮驴一年成绩"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基于小蛮驴的技术底盘，达摩院正在开发多种自动驾驶产品，如具备自主移动能力的电力巡检机器人。首批量产的巡检机器人，以服务机器人的成本实现了工业机器人的性能，即将批量进入电力系统，取代人工开展电力巡检工作。"}]}]}