Global and China L4 Autonomous Driving Industry Report, 2019-2020

Giants gain high finance.
Progress of L4 autonomous driving is greatly hampered over the recent two years, causing OEMs’ and Tier 1 suppliers’ delay in L4 launches. Yet, the top L4 companies still raised huge funds in the past year.

In 2019, Baidu, Pony.ai and WeRide succeeded in commercial pilot of Robotaxi on complex urban roads in limited areas, a crucial step for L4 autonomous driving in China.

L4 autonomous driving technologies starts to find real application but gets deployed by most players first in one scenario or two as the current L4 cannot be perfectly suited to all driving scenarios.

It is shown from the planning of the OEMs and the providers of technical solutions for autonomous driving that L4 autonomous driving will be commercialized in limited scenarios ahead of open scenarios in the following three to five years.

Baidu, UISEE, DeepBlue and Trunk Tech all have conducted L4 trials in limited scenarios such as parks, ports, and airports. Besides, Baidu, Momenta, Bosch, ZongMu and UISEE are vigorously deploying in the parking lot scenario.

Deployments in open scenarios cover Robotaxi on urban roads and autonomous trucks on the expressways. In the Robotaxi field, Waymo, Baidu, Pony.ai, and WeRide have carried out pilot projects in both China and the U.S. in specific areas of a city, manned by safety officers, and WeRide is already fully open to the public in Nov.2019. Software & hardware technology iteration and larger-scale tests are essential for open Robotaxi in wider areas.

Technically, the current L4 autonomous vehicles for tests (or trial operation) are largely provided with the single-car intelligence solution. For safer autonomous driving, solution providers make vehicle’s environment perception capability optimized ever. For instance, Voyage’s G2 autonomous vehicle carries Velodyne's VLS-128 LiDAR system with a detection range up to 300 meters, tripling the capability of the 64-channel LiDAR installed on the 1st-Gen autonomous vehicle. Boasting 29 cameras, Waymo’s next-generation autonomous driving system enables a detection range of 500 meters whist improving LiDARs’ performance significantly.

Also, it is of vital importance to enhance vehicle’s motion perception competence. High-precision positioning can be realized by integrating high-precision positioning modules (composed of 5G modules, IMU and HD maps).

In the short term, either single-car intelligence solution or strong field terminal solution fulfills L4 autonomous driving in the confined areas, but in the long run, collaborative vehicle infrastructure system (CVIS) is the mainstream technology roadmap for L4 autonomous driving. Through CVIS, the vehicle is fully connected to the “X” as spatiotemporal dynamic traffic information are collected and integrated whilst the active safety control of vehicle and the collaborative management on roads are done for safer running of autonomous vehicle.

By analyzing the test data, Baidu concludes that CVIS can solve 54% of the problems encountered in road tests and 62% of the takeovers incurred by single-car intelligence, adding redundancy for autonomous driving safety.