Urban intelligent driving, perhaps this is the closest we will get to it.

In 1985, for Irwin Jacobs, it might have been the worst year or the best year.

In this year, he reluctantly left the 17-year-old Linkabit due to conflicts with investors, and in this year, at the age of 52, he was determined to create Quality Communication.

So, taking the first letters of the two words, Qualcomm was formed, now the largest mobile chip manufacturer in the world.

The fledgling Qualcomm staked everything on CDMA technology. However, the new technology was not immune to the problem of being forgotten, and the cancellation of the only customer became the critical straw.

As the straw fell, Qualcomm chose to transform the project into Omnitracs, a satellite communications and positioning system for long-distance trucks. This system solved the truck management problems of freight companies and quickly became a hit in the logistics industry, and Qualcomm took advantage of the opportunity to catch its breath.

It can be said that Qualcomm and the automotive industry had a fateful relationship from the beginning, and it has continued into 2022.

The first light shone in 2022.

The first seven months of 2022 may have been the most anticipated or the most difficult to wait for in the Chinese intelligent driving market.

In this year and a half, the hardware capabilities of intelligent vehicles have been qualitatively improved. New energy vehicle models equipped with one or even multiple lidars have been released and put on the market, and lidar is seen as the key weapon to open the next stage of intelligent driving.

In this year and a half, the software capabilities of intelligent vehicles seem to be still lagging behind. Despite greater computing power and lidar, the higher-order intelligent driving functions that people enjoy are still mostly in planning and PPTs.# City NOH Takes the Lead with the Release of the Mocha DHT-PHEV LIDAR version powered by Snapdragon Ride Platform

As a concrete representation of advanced intelligence driving functions, the smart assisted driving systems of cities in China are like iron walls that everyone is awaiting an opportunity to crack and bring to light. Will it be Tesla, with their rich experience from abroad? Will it be Xpeng, the Chinese smart driving leader? Or will it be NIO, whose overall computing capability has exceeded expectations?

Perhaps no one expected that the first ray of light would come from the relatively traditional Chinese car brand, Great Wall Motor, with its innovative Weyp brand. On August 26th, at the Chengdu Auto Show, Weyp officially released the Mocha DHT-PHEV LIDAR version, declaring that City NOH has taken the lead in landing applications.

The reason why it is dubbed a surprise is because they have connected the “traditional” with “leading the way”. And if we have a pair of eyes that can see through the underlying details of this car, we will find that the key core of the connection is marked with a red and white circular logo – the Qualcomm Snapdragon Ride platform, which also made its first appearance in a mass-produced car model in China.

The Snapdragon Ride platform, which is one of the four components of the Snapdragon Digital Chassis, also carries all of Qualcomm’s ambitions in the era of intelligent driving. It is aimed at advanced parking and overall integration.

At the Qualcomm Snapdragon Tech Summit in December last year, the Snapdragon Digital Chassis and the Snapdragon Ride platform were unveiled. The former occupied one-third of the time at this year’s Qualcomm Snapdragon Tech Summit, while the latter was pre-announced by Qualcomm’s Vice President, Hou Mingjuan, five months ago stating that “this year is the first year for Snapdragon intelligent driving technology to go on board.”

The first move to go on board the Snapdragon Ride platform was to release the City NOH “joker card”.

The Deduction of Light

So, how was the “joker card” shown?

Not long ago, Qualcomm and Weyp jointly released the on-road test video of the Weyp Mocha DHT-PHEV LIDAR version, a sneak peek for the upcoming Q4 delivery season.“`markdown

Overall, Snapdragon Ride is an advanced intelligent driving assistance solution that supports “city + highway” scenarios such as highways, open city roads, city expressways, and parking lots.

Getting into the details, the Mocha DHT-PHEV LiDAR version in the video encounters various types of corner cases in large-scale scenes, such as being able to handle dozens of tunnels, adapting to hundreds of road types, driving smoothly in extreme cold weather, and even going to high altitude areas.

In these large-scale scenes, we usually only encounter a part of the first two types; therefore, this is more like a durability and stability test for Qualcomm.

And the focus that everyone is more concerned about is naturally the advanced intelligent driving assistance on city roads. Here are a few that I am interested in:

  • Lane change avoidance“: When a vehicle in an adjacent lane cuts in, the display of the SR environment simulation has already completed the target switching of the following vehicle as soon as both wheels of the passing vehicle cross the lane line. When half of the width of the cutting-in vehicle enters, it is automatically marked as red because of insufficient longitudinal clearance. After stable recognition, the subsequent braking and avoiding actions will appear very natural.

  • Traffic light recognition“: This feature is not uncommon in itself, but when paired with intelligent navigation assistance in China, drivers no longer have to worry about any strange behaviors when driving with intelligent driving assistance and being the first car at the red light.


“`- “Unprotected left turn“: When the vehicle detects a pedestrian, it does not take a sudden panic brake, but generates a linearly increasing braking force based on the distance between the vehicle and the person, showing a human-like manner.

  • Intersection game“: The vehicle monitors both the signal lights at the intersection and the status of oncoming vehicles. Even if the oncoming vehicle has already approached, the intelligent driving state does not exit, but avoids the vehicle first and then deals with the yellow light.

  • Night driving“: The advantages of LiDAR are fully demonstrated. With LiDAR in hand, no matter how dark the sky is, vehicles in the same direction, oncoming vehicles, guardrails, street lamps, pedestrians, non-motorized vehicles, and even roadside trees are clearly visible in the point cloud image.

In addition to these, the video actually listed 9 scenes on one screen. They may not be so “corner” but they are almost all cases we encounter in our daily driving.

The Secret of Deduction

Speaking of advanced intelligent assisted driving in the city, China is not short of related players, but why did the thunderous noise of previous press conferences and demos result in such difficulty in the mass production process?

He XPeng, the helmsman of XPeng Motors, revealed the truth behind it in an interview, “When we were about to release the NGP, we thought the city NGP would be faster, but in fact, the city NGP is not that fast. The biggest difficulty lies in the management of high-precision maps.” The maps he mentioned actually refers to high-precision maps.

High-precision maps can be understood as an advanced version of ordinary maps, which also includes hundreds of additional information such as lanes, traffic lights, guardrails, street lamps, etc.The essence of environmental sensing in autonomous driving is actually information acquisition. According to research institutions, the complexity of urban road conditions in China is 15 times that of California in the United States. The fundamental solution to solve this contradiction of complex road conditions is to obtain more information. Compared with perception, high-precision maps that are pre-drawn can significantly reduce the perception pressure of intelligent driving hardware. This is why domestic intelligent driving players have focused on multi-sensor fusion + high-precision map solutions before this year.

However, the precision of high-precision maps brings more severe “double-edged sword” effects. On the one hand, it is indeed a stable information source for intelligent driving systems; on the other hand, the massive data pre-acquisition means higher costs and longer production and update cycles. At the same time, overly detailed data also brings risks related to national defense and security. Therefore, high-precision maps have become a type of technology subject to strong supervision and approval.

The key to Qualcomm Snapdragon Ride’s ability to lead the industry is that they have taken a new path between “pure visual perception” and “multi-sensor fusion + high-precision maps”, that is “heavy perception + light maps”.

The first step to achieving “heavy perception” is more extensive environmental perception hardware.

The Mocha DHT-PHEV LiDAR version is equipped with 2 125-line LiDARs, 5 millimeter-wave radars, 12 ultrasonic radars, and 12 cameras to construct a fully covered perception defense line around the vehicle.

But “seeing” is only the process, “understanding” is the goal. How to identify, understand, and construct a vehicle’s surrounding environment model based on massive perception data? The Qualcomm Snapdragon Ride platform surpasses all others in this regard.As the first 5nm ADAS and autonomous driving solution in the automotive industry, and currently the highest individual computing power among mass-produced chips, with 360TOPS computing power in hand, it has application processors paired with autonomous driving accelerators on the hardware side. Qualcomm has loaded the concept of mobile processors onto automotive chips, coupled with its expertise in high-performance, low-power consumption computing and flexible platform combinations, achieving scalability rarely seen in the automotive industry. In terms of software, it provides a comprehensive and complete software stack, giving chip developers the key functions of intelligent driving, such as the fusion of positioning and maps, perception and sensors, and behavior prediction and planning, in a user-friendly manner.

And these difficult technical terms have created the first landing of City NOH, which in turn proves the correctness of the technical route chosen by the Snapdragon Ride platform.

That is to say, “heavy sensing, light mapping, and large computing power” are the main theme of the next stage of development of intelligent driving.

And the main theme often attracts more singers, not just Wei Pai, but also BMW, GM, and Volkswagen… Qualcomm provides a comprehensive set of development tools for its partners, and the “Snapdragon Smart Driving” of these giants is already on its way to us.

How familiar is this scene?

34 years ago, Qualcomm had an accidental encounter with the automotive industry; 20 years ago, Qualcomm officially entered the automotive industry by providing 2G connections for cars. Seven years ago, Qualcomm began researching and developing autonomous driving; one year ago, Qualcomm’s intelligent cockpit chip 8155 began to become the standard equipment for in-car “God u”.

So, how far away is Snapdragon Ride from being the standard for full-scenario intelligent driving?

-END-

This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.