The Chinese search engine company Baidu has won permission to test its self-driving cars in Beijing. China's capital Thursday issued its first temporary license plates for testing on the city's public roads to Baidu's self-driving vehicles.


A Baidu self-driving car uses the company's Apollo system. (Photo courtesy Baidu)


Beijing opened 33 roads of a total length of 105 kilometers (65 miles) for autonomous car testing outside the Fifth Ring Road on the outskirts, in the Haidian District, away from crowded areas but near the Baidu campus.


Baidu is developing high-resolution maps for self-driving cars; the first map is to be based on the 33 roads.


The licenses for Beijing road testing are valid for only 30 days, but license holders can apply for renewal after the self-driving cars pass government assessments.


The autonomous test vehicles must be equipped with devices that can monitor their location, can monitor whether or not a test car is in self-driving mode, and the behavior of each of the vehicles, and test drivers for the autonomous cars must have taken at least 50 hours of training in handling self-driving cars in case they have to take over.


In April 2017, Baidu launched its Apollo project, a self-driving vehicle that includes the vehicle platform, hardware platform, software platform and cloud data services.


Baidu claims the Apollo car has the world's first human-car AI interaction system. It offers human-vehicle dialogue, face registration, fatigue monitoring, augmented reality navigation, smart security and vehicle to home (V2H) connectivity.


Sensors, such as LiDAR, cameras and radar collect environmental data surrounding the vehicle. Using sensor fusion technology perception, algorithms can determine in real time the type, location, velocity and orientation of objects on the road, as required by law.


This autonomous perception system is backed by Baidu’s big data and deep learning technologies, as well as a vast collection of real world labeled driving data. The large-scale deep-learning platform and Graphics Processing Unit (GPU) clusters shorten the system's learning time for large quantities of data.


Once trained, the new models are deployed onto the vehicle using over-the-air updates through the cloud.


Apollo’s perception system can continuously improve its detection and recognition capabilities, which provide accurate, stable, and reliable input for other autonomous system modules.


Baidu is working to create more cost effective sensors as well integrate Apollo into a wider variety of vehicles including minibuses, SUVs and pickups.


There are multiple sets of sensors in the vehicles so that, in case of sensor failure, collisions with pedestrians or other vehicles can be avoided, an assurance that's especially important following a fatal collision of an autonomous car with a pedestrian walking her bike in Tempe, Arizona.


Baidu plans to gradually introduce fully autonomous driving capabilities on highways and open city roads by 2020.


China issued licenses to automakers allowing self-driving vehicles to be road tested in Shanghai earlier this month. Shanghai-based SAIC Motor Corp Ltd. and electric vehicle start-up NIO received licenses.


China's regulations managing road testing for self-driving vehicles say that autonomous vehicles are eligible for public road testing only after they have completed 5,000 kilometers (3,000 miles) of daily driving in designated closed test fields and passed government assessments.



By Sunny Lewis

Environment News Service (ENS)