Autonomous driving may have nothing to do with artificial intelligence for a long time

Autonomous driving may have nothing to do with artificial intelligence for a long time

People in the Internet industry like to talk about trends, and they like to put trends together, as if this can create an effect where 1 plus 1 is greater than 2.

Moreover, some of the technologies that are highly anticipated may indeed have intrinsic connections, so they are more likely to be treated as synergistic industries that stimulate each other, and it is expected that the synergistic explosion will produce earth-shaking effects.

For example, artificial intelligence and driverless cars.

Indeed, it is undoubtedly inevitable that humans will use artificial intelligence technology to handle tasks such as mechanical operation, driving, and surveying.

But inevitability does not mean immediate realization. We are now at the beginning of the development and commercialization of many technologies. From 0 to 10, 1 and 9 are indispensable. Currently, artificial intelligence and driverless cars may be in such a relationship.

The first level of autonomous driving must be low-speed and fixed-scene applications

First of all, we need to know what autonomous driving is.

The core judgment point of unmanned driving lies in breaking away from the human efforts of autonomous driving and the operation methods of traditional cars, and turning the vehicle into an entire driving process that actively completes the entire driving process according to the set goals.

From the perspective of technological maturity and application possibilities, driverless cars can never be achieved overnight and can be driven directly on the complex road conditions of urban traffic.

Due to concerns about safety and the need for data accumulation, there are two constraints on early driverless cars:

The first is speed. When autonomous driving is first put into use, it will not be possible to drive at high speeds like normal vehicles. Instead, it must start with low-speed vehicles below 50 kilometers per hour, and gradually test and improve the reliability of the autonomous driving system.

The second is the application scenario. It is basically impossible to put unmanned driving into use in complex road conditions immediately in the initial stage. We are more likely to see unmanned vehicles used in campuses, scenic spots, corporate parks, and airports. On the one hand, the road conditions in these scenarios are simple, there are few problems to be dealt with by unmanned driving, and unmanned vehicles can be replaced as a whole to ensure the feasibility of interaction between vehicles. On the other hand, these scenarios have low requirements for transportation capacity and short usage time, thus meeting the characteristics of low speed and strong monitoring of unmanned driving in the initial stage.

Therefore, in the early stage of driverless driving, low-speed electric shuttles will definitely be eliminated, and at this stage, the involvement of artificial intelligence is not that strong.

Compared with deep learning, risk avoidance, precise driving and human-machine interaction are more important.

Artificial intelligence manages driverless vehicles. The logic is to solve various problems encountered during vehicle driving and ensure driving safety through deep learning and response mechanism analysis system.

However, the problems that vehicles encounter in real traffic scenarios are ever-changing, and each problem may be a new one. Many of them are difficult for law enforcement officers to analyze, not to mention the artificial intelligence system based on past case studies. Especially in the selection of traffic plans, in complex and congested road conditions, it is difficult for current artificial intelligence algorithms to completely replace experience and human judgment.

In low-speed unmanned driving, there are not enough application points for artificial intelligence. In this scenario, there are three main technologies to ensure driving safety and smooth operation: risk avoidance technology, precision driving technology and new human-machine interaction.

Risk avoidance is a premonition factor to ensure the safety of driverless vehicles. At present, the best solution is to use hardware to judge the distance between the vehicle and other objects and the possible collision speed, and then command the driving system to complete the avoidance. Therefore, perception hardware is the core.

Precision driving is a technology and hardware system that ensures that vehicles can achieve their transportation goals. It mainly involves route determination based on maps, GPS, and vehicle perception. In addition, smart speed change and terrain modules are also important.

What is often overlooked is the human-vehicle interaction in the unmanned driving mode. Under the autonomous driving technology, the human-vehicle interaction mode has not changed much. However, in the unmanned driving mode, the human command mode needs to change from full operation to emergency command and order issuance, and a full set of operating systems are needed to adapt to this demand.

Looking back at these driverless car demands, we will find that it is currently difficult for artificial intelligence to distinguish any of them. Therefore, artificial intelligence in the early stages is more likely to join the driverless car technology group as an auxiliary technology rather than truly becoming a central commander.

Of course, artificial intelligence will definitely become the way humans complete transportation in the future, but that requires a substantial evolution of artificial intelligence itself and the improvement of the underlying technology and hardware of manual driving. At present, there is still some distance before the two meet unexpectedly.

These technologies are the most important for primary driverless driving

So let’s take a look at which technologies are the most important for primary autonomous driving and which are most likely to change the speed at which autonomous driving enters the market.

1. Lidar. Radar sensing the external environment seems to be the most mature driverless technology at present. Moreover, companies such as Google Waymo have announced that they will provide new solutions on Lidar hardware, which will significantly reduce the cost of Lidar. It is very likely that new large hardware service providers will emerge in this field and quickly invest in the configuration of autonomous driving vehicles to welcome the arrival of driverless driving.

2. Multi-sensors. Sensing and mapping systems are the core of solving the potential safety problems of unmanned driving, especially in avoiding risks in complex environments. There are many solutions for unmanned driving sensors, and the most likely to occupy the market in the end is the multi-sensor that combines sound waves, infrared rays, and thermal energy.

3. Visual camera. The camera is the eyes of the driverless vehicle and the basis of the driverless vehicle's sense of space. However, the visual processing system and application hardware linked to the processor have not been particularly mature. There are many new technologies in this field, and artificial intelligence can also be added to assist. It should be the highlight of the next driverless car.

4. Ultra-precise map system. Rather than letting driverless cars calculate routes intelligently, a better solution is to let all vehicles move according to the map to maximize transportation capacity and traffic safety. This requires real-time maps with decimeter accuracy. This is also the biggest potential point for domestic and foreign map companies to combine with driverless cars.

5. Human-computer interaction system. As mentioned before, the human-computer interaction system for unmanned driving will be brand new. Then the specific application methods and standardization will become the key technologies that are urgently needed. At present, there are few new technologies in this field, but the entry threshold is not high, which is suitable for entrepreneurs to go deep.

6. Inter-vehicle interaction system. Based on the Internet of Vehicles system, vehicles can sense each other and avoid each other, which is a direct response of the Internet of Things technology in the field of unmanned driving. Complete vehicle interaction should be able to achieve comprehensive interaction between vehicles and people, vehicles and vehicles, vehicles and roads, and vehicles and networks. This direction is the core breakthrough of technology.

Through these urgently needed technical breakthroughs, it is not difficult to see that the current demand for hardware, software, and systems for autonomous driving is very strong. The successive outbreak of these technologies and the attention of capital will be the norm in the autonomous driving market in the future. Compared with the love of small technologies and small hardware by American and Israeli teams, domestic players seem to be too obsessed with vehicle manufacturing and artificial intelligence. In the end, it is very likely that the core technology will be in the hands of others again, and even the expected big outlet is too far away and will not come.

The big future comes from small breakthroughs. At least for now, it is better not to think of autonomous driving and artificial intelligence so closely.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.

<<:  It's not true that Tesla can predict accidents

>>:  Is Toutiao worth $10 billion?

Recommend

Analysis of Pop Mart’s Private Domain Operations

The blind box craze has made Pop Mart popular, ma...

Operation and promotion plan after the event goes online!

First look at this picture: I divided the operati...

Successfully launched! All members of Gaofen-3 have assembled!

The reporter learned from the National Space Admi...

9 factors that content marketing veterans won’t ignore when planning

Effective marketing increasingly relies on storyt...

How much does it cost to customize a skin care product mini app in Hezhou?

According to industry insiders, mini programs wil...

How to quickly increase the volume of corporate short video operations?

At that time, some friends said that it was due t...