US chipmaker Nvidia has teamed up with Chinese ride-hailing giant Didi Chuxing to develop autonomous vehicles for a scalable ride-hailing service, as global companies join forces to accelerate autonomous car deployment.
Why it matters: Due to the immense amount of computing power needed for autonomous driving, automakers and mobility services have been seeking out partnerships with chip makers.
- The partnership echoes an earlier deal between Nvidia and Toyota, a major Didi backer, in which Toyota will use the chipmaker’s platform to test, validate, and deploy autonomous cars to the mass market.
- James Kuffner, chief of Toyota Research Institute-Advanced Development, told reporters on Tuesday that the company plans to first develop and deploy self-driving technologies in commercial vehicles for services including ride-hailing before producing highly autonomous passenger vehicles.
Details: Didi has selected Nvidia Drive, an end-to-end computing platform to develop, train, and validate its driverless technologies, Nvidia CEO Jensen Huang announced at its graphics processing unit (GPU) conference in the eastern Chinese city of Suzhou on Wednesday.
- Didi will enhance its autonomous driving technologies by using Nvidia’s platform to train deep neural networks, which powers a self-driving car with visuals of its surrounding environment and help it make decisions based on what it sees.
- The ride-hailing giant will also use Nvidia’s GPUs in the data center for training machine-learning algorithms, which the company uses for route planning more than 40 billion times each day.
- It follows an announcement Didi made in August of plans to launch a robotaxi service pilot with a fleet of 30 L4 self-driving vehicles in the outskirt of Shanghai early next year.
- The city government has yet to grant permission for a passenger transport program, people close to the matter told TechNode earlier this month.
Context: Robotaxis are seen as the most likely business application for self-driving technology given the high costs and strict regulations required to mass produce autonomous cars for personal use.
- Nvidia’s top competitor, Intel, also joined forces with an electric vehicle maker when it announced last month a partnership with Nio to release a highly automated model in China in 2022.
- Amnon Shashua, CEO of Intel’s automotive sensor company Mobileye, expressed a positive view about a rollout in China due to its centralized regulatory environment, adding that Chinese regulators were currently standardizing Mobileye’s safety model into law.