The game between GM and Tesla in the automatic driving safety mechanism

- Jul 24, 2018-

In the long-distance running of automakers in automakers, GM hopes to grab a seat with Cadillac. According to GM's plan, starting from 2020, the semi-automatic driving system known as Super Cruise will be installed on all Cadillac models. This is GM's desire to challenge Tesla's dominance in the mass-production autopilot market share.

Compared to Tesla, although super cruising is also a technology that does not require the driver to control the steering wheel, GM recognizes the need to shift drivers from relying solely on autonomous driving to more attention to driver awareness. Unlike fully autonomous driving techniques in other vehicles, Super Cruise includes a highly demanding safety code to ensure that the driver maintains a focus on the vehicle's surroundings, even if the vehicle is not directly controlled.

Currently, Super Cruise is primarily open to highway autopilot systems. After the high speed is turned on, the speed is controlled within the speed limit range, and then the direction is kept steady. After a few seconds, the car will judge that it has met the Super Cruise open condition and give a green steering wheel icon on the LCD instrument panel. Go to this icon and press the Super Cruise button on the steering wheel to enter the Super Cruise mode and then leave the steering wheel with both hands.

How to ensure absolute safety, the general approach is to increase the driver's attention to maintain the system. There are more warning lights on the steering wheel, and two more cameras that capture the driver's eyes and face. The two cameras work together. If the system determines that the driver's attention is not concentrated in front, it will send a reminder through the light strip on the steering wheel and other means.

Earlier, Tesla CEO Elon Musk has been arguing that he refused to carry the corresponding driver monitoring function on Tesla models. In recent months, driver monitoring has become a new urgency, especially as more and more car OEMs begin to accelerate the deployment of L2-class autopilot functions.

As a definition of L2 level autonomous driving, the driver is still responsible for the safe operation of the vehicle. However, under certain conditions, the car can turn on the steering, braking and acceleration functions. The driver still needs to do everything else, including monitoring the road conditions. Currently, globally mass-produced systems include GM's Super Cruise, Mercedes-Benz's Drive Pilot, Tesla's Autopilot, Volvo's Pilot Assist and Nissan's ProPilot Assist.

The National Transportation Safety Board investigated the lack of adequate monitoring in the Tesla report, a survey of a fatal accident involving autonomous pilots in Florida in May 2016. The driver's lack of concentration during the Uber crash in Arizona earlier this year also highlighted the importance of autopilot safety mechanisms, including driver monitoring.

If, next, GM's Super Cruise is proven to be the driving accuracy and timeliness of driving awareness and autonomous driving technology, this super-cruise system is likely to become the standard of the future industry, providing more for autonomous driving technology. Production experience.

Of course, Tesla is not "stubborn". Recently, the owner of the Tesla Model 3 found a very hidden camera in the cab. This camera is currently off, and may provide hardware support for driver's attention monitoring in autonomous or semi-automatic driving in the future, but Tesla is still tight-lipped about this plan.

Tesla also confirmed that in addition to the driver's surveillance within the camera, the entire passenger compartment, including the front and rear rows, can be covered by its range.

But the driver monitoring system is not a panacea. Since last year, the National Highway Traffic Safety Administration has shown that many systems still cannot effectively grasp the driver's distraction and distraction.

Cadillac's "automatic driving" is also super cruise, which is regarded as the advanced assisted driving function of ADAS. It can not be counted as automatic driving. He has made certain premise restrictions in the scope of application.

But then, the future of GM's autonomous driving program will be largely embraced by consumers' current production-grade super-cruise systems, but consumer acceptance may make GM a long-term upgrade to autonomous driving technology. power.

Whether super-cruising technology will become an industry-wide safety standard remains to be determined, but it will certainly affect the short-term future of autonomous driving technology and the way consumers drive their vehicles.

Nowadays, the world's major automobile OEMs are fighting in two lines. On the one hand, they promote the commercialization of L4-class autopilot by means of shared travel/unmanned taxis, and on the other hand, they are constantly iterating ADAS for ordinary consumers. Mass production of L3-class autonomous driving models.

Objectively speaking, the GM Super Cruise system is not perfect, and even more people think that this is just a more advanced ACC+ lane keeping. But Super Cruise has a lot of redundant configuration to increase the safety of semi-automatic driving. So far, this is the best landing business model.

After all, the L2-capable models currently on the market are still more of a “show off” function, rather than a complete set of automatic driving safety mechanisms.

But there is still a problem that still needs to be resolved, that is, user privacy.

Previously, the US consumer report evaluated 44 brand models, including 32 models with built-in wireless data connectivity, in addition to the car camera and in-vehicle surveillance camera. As mentioned in the Tesla Model3 manual mentioned above, the in-cabin camera is not yet open, but it can be used for future potential functions. At the same time Tesla promised that they would let customers know that they turned on the camera.

For GM's super cruise system, the company previously promised not to capture or store video. But with the trend of more cameras and sensors in the vehicle configuration, the car is collecting a variety of data on driving habits. Of course, much of this data is collected to help automakers test autopilot technology, which is considered a machine learning process.

At the same time, the current global legal rules surrounding automotive data are vague. This may be a potential risk challenge that autopilot safety mechanisms need to overcome.