DeepRoute-Driver 2.0 currently costs less than $10,000, and the cost of the self-driving system could be reduced by another 70 percent after large-scale partnerships with automakers.
(Image credit: DeepRoute)
For self-driving fleet operators, balancing technology and cost is key to achieving commercial success. Alibaba-backed DeepRoute appears to be making significant progress in this regard.
DeepRoute today unveiled its first fleet of L4 self-driving front-loaded solutions, consisting of 30 vehicles that will be put into the company's Robotaxi operation in Shenzhen.
The fleet features DeepRoute's latest L4 autonomous driving solution for front-load mass production, DeepRoute-Driver 2.0, which is currently the lowest cost L4 autonomous driving solution in the industry, the company said.
In addition to creating safe and reliable technology, there are many issues that need to be addressed to achieve mass production of autonomous driving. Among them, how to achieve cost reduction and efficiency is the most practical issue, according to DeepRoute CEO Zhou Guang.
DeepRoute-Driver 2.0 currently costs less than $10,000, and after large-scale cooperation with car companies, the cost of the entire autonomous driving system can be reduced by another 70 percent, he said.
DeepRoute announced the L4 self-driving solution on December 8 last year, saying it can help vehicles drive themselves on complex city roads and perform better than Tesla's FSD (Full Self-Driving).
Best of all, it costs less than $10,000, nearly the same price as FSD and the lowest recorded in the industry, DeepRoute previously said.
DeepRoute-Driver 2.0 uses five solid-state LiDARs and eight cameras for 360° sensing. Its LiDAR suppliers are RoboSense and Zvision.
It uses sensors that are mostly embedded, allowing self-driving vehicles to eliminate the "hat" on top of the vehicle, according to the company.
DeepRoute-Driver 2.0 has been in the Shenzhen CBD for six months of regular road testing, and can complete complex operations such as changing lanes and avoiding temporary roadblocks with ease, the company said today.
The solution, which can be adapted to a wide range of vehicles, has been successfully adapted to Rising Auto's MARVEL R, Lincoln MKZ, Geely Geometry A and other models, allowing one-touch access to the autonomous driving system, it said.
DeepRoute-Driver 2.0 uses the Nvidia DRIVE Orin chip, and the company's inference engine technology keeps the overall power consumption of the computing platform to 150 watts, DeepRoute said.
DeepRoute-Driver 2.0 has an accurate sensing range of more than 200 meters, the company said, adding that to ensure safety, it has designed a redundant vision-only sensing system that adaptively drives itself according to safety policies even in the event of partial sensor failure.
DeepRoute has already begun working with a number of car companies on the technology. Cars equipped with the company's L4 autonomous driving solution are expected to begin mass production and enter the market on a large scale by 2024, according to the company.
Vehicles will then be the first to achieve L4 autonomous driving in areas with rich data accumulation, and continue to accumulate data in other areas in a human-machine co-driving mode first, before gradually releasing L4-level autonomous driving capabilities and eventually achieving autonomous driving in more areas, it said.