New Solution for Assisted Driving: Sensing Takes Precedence Over Mapping

With the continuous iteration of technology, assisted driving functions of electric vehicles have entered a higher stage of competition. At present, there are two main types of urban navigation assisted driving systems in China. One of them is the “multi-sensor fusion + high-precision map” scheme used by domestic automotive manufacturers such as XPeng, while the other is the “sensing takes precedence over mapping” approach proposed by self-driving technology company Haomo.AI. Currently, the two schools are neck and neck, according to the National Business Daily‘s report.

In addition to larger space and a 800V high-voltage supercharging platform, one important selling point of the XPeng G9, which will be launched on September 21, is that it will realize intelligent assisted driving functions for urban scenarios. Haomo.AI, which was incubated by Great Wall Motor, said on September 13 that the Mocca DHT-PHEV LiDAR version with its urban Navigation on HPilot (NOH) function will go into mass production in September and will go on sale later this year. Although both models have urban assisted driving features, XPeng and Haomo.AI take diametrically opposite technological routes.

SEE ALSO: China’s Smart Car Sector Now Ahead of Europe and US, Haomo Chairman Says

The three elements of autonomous driving technology are perception, decision and control, among which the first two cannot be separated from the support of high-precision maps. Compared with ordinary navigation maps, high-precision maps have greater accuracy, more data dimensions and more accurate positioning.

At present, NIO‘s ES8, ES6, and EC6 all use Baidu‘s high-precision mapping to realize auxiliary driving functions including autonomous lane change and on-ramp, while GAC Aion LX’s ADiGO 3.0 system also uses Baidu‘s high-precision map to realize autonomous driving on highways. XPeng‘s Navigation Guided Pilot (NGP) system uses the high-precision map provided by AutoNavi’s third-generation vehicle navigation.

However, current high-precision maps have limitations, including slow updates and high collection costs. Meanwhile, there are also problems with the format exchange specifications of data maps, data interconnection standards, and connection with vehicles.

A notice issued by China’s State Ministry of Natural Resources on August 31 sets out clear rules and requirements for the production of high-precision maps. The mapping and drawing of high-precision maps can only be legally operated by enterprises that have first-class mapping qualifications for making navigation electronic maps. This means that the qualifications for making a high-precision map are tightening.

At this stage, not relying on high-precision maps has become a realistic demand of car and technology companies when developing advanced assisted driving functions. Richard Yu, CEO of Huawei’s Terminal BG and Intelligent Vehicle Solution BU, once publicly stated that high-precision maps are still used in autonomous driving, but that future development should not rely on them.

In addition, Tesla CEO Elon Musk has repeatedly said that high-precision maps are a “terrible idea” and that the autopilot system will lose the flexibility to respond like a real driver if it relies too much on stored high-precision maps.

In this context, autonomous driving technology startup Haomo.AI and other enterprises have put forward the route scheme of “sensing takes precedence over mapping.” “With perceptive technology, urban assisted driving can be achieved without the need for extremely accurate maps,” Pan Xing, the technical director of Haomo.AI, told the National Business Daily. Of course, this will mean more lidars mounted on vehicles.

Analysts believe that there are two advantages of not relying on high-precision maps: avoiding the limitations of policy approval and cutting costs.