For a long time, many industry organizations, consulting companies and main engine factories have believed that it is difficult for China to realize autonomous driving at this stage, for no other reason, because China's road conditions are too complex, the sudden emergence of electric bikes at intersections is likely to surprise the autonomous vehicles and cause traffic accidents. But the problem is likely to be solved by Cruise, gm's autonomous driving company, which has developed a system called Continuous Learning Machine that predicts dangerous situations on the road.

According to Cruise, Continuous Learning Machine can continuously "train" itself through the superposition of driving distance, enabling its AI to recognize and analyze the intentions of other drivers and pedestrians on the road. For example, the system senses that a freshly parked car is likely to open its doors suddenly, or reverse, turn, etc.

From the perspective of working principle, Continuous Learning Machine system can record data in vehicle running. When the prediction made by it is quite different from the actual situation, this scenario will be added to the training data set. The system then learns the scenarios where the data is concentrated to improve autopilot safety. That way, even the u-turns and K-u-turns, which are more difficult for drivers to handle, won't faze Mr. Cruise's autonomous vehicles.

Cruise is understood to have launched Origin, its first autonomous vehicle, in early 2020. With no brake, accelerator or steering wheel, the car has room for only six passengers, a start button, a mayday button and an internal camera, making it a truly "driverless" vehicle. With a Continuous Learning Machine, a system of autonomous Learning that can predict road extremes, Origin may be able to enter the "home of the average person" much earlier.