Beijing (Gasgoo)- On March 23rd, China’s autonomous driving startup, DeepRoute.ai announced that the company has adopted NVIDIA’s Drive Orin systematic chip for its production-oriented L4 autonomous driving solutions.
Photo credit: DeepRoute.ai
With its self-developed inference engine technology, DeepRoute.ai is able to optimize its computing resource usage by integrating its algorithms with the Orin chip. The Orin chip has a maximum of 254TOPS computing power, helping DeepRoute.ai lower the production cost of its L4 autonomous driving solution. With two pieces of the Orin chip, the energy consumption of the solution’s computing platform was reduced to only 150W, compatible with various autonomous vehicle models.
In fact, DeepRoute.ai has maintained a profound relationship with NVIDIA since 2019. In early 2020, the company has integrated the NVIDIA Xavier in its low-consumption computing platform, and successfully slashed the platform’s cost to half of traditional solutions’ cost.
Software-wise, DeepRoute.ai studied the features of different chips on the market and developed an inference engine technology. The technology allows its autonomous driving algorithms to run smoothly and stably on its low-cost, low-consumption hardware platform while gaining six times the inference speed of mainstream deep learning frameworks.
According to the CEO of DeepRoute.ai, Zhou Guang, the company will put NVIDIA’s Drive Orin systematic chip on its upcoming Robotaxis in 2022, and conduct scaled road tests.
