A driverless car is a type of smart car, also known as a wheeled mobile robot, which relies mainly on a computer-based smart pilot in the car to achieve the goal of unmanned driving.
According to Thomson Reuters' latest report on intellectual property and technology, between 2010 and 2015, more than 22,000 invention patents related to automotive driverless technology, and in the process, some companies have emerged as an industry in this field. leader.
A driverless car is a smart car that senses the road environment through an in-vehicle sensing system, automatically plans driving routes, and controls the vehicle to reach a predetermined target.
It uses the on-board sensor to sense the surrounding environment of the vehicle, and controls the steering and speed of the vehicle based on the road, vehicle position and obstacle information obtained by the perception, so that the vehicle can travel safely and reliably on the road.
It integrates many technologies such as automatic control, architecture, artificial intelligence and visual computing. It is a product of high development of computer science, pattern recognition and intelligent control technology. It is also an important indicator to measure the scientific research strength and industrial level of a country. The national economy has broad application prospects.
Which is the driverless technology?The latest list released by foreign media shows that Google’s autopilot Waymo ranked first in the overall score. Baidu from China ranked seventh, and Didi became the lowest score among the 17 companies in the driverless field. On June 13th, the US technology media The InformaTIon conducted a comprehensive evaluation and ranking of 17 major companies developing L4 and above unmanned technology through interviews and research.
Among them, Waymo, the parent company of Google's parent company Alphabet, scored the first with 13 points, and Uber ranked second with 1 point. German old car manufacturer Daimler and American auto parts supplier Delphi were tied. Third place.
Foreign media ranking autopilot technology: Google wins Baidu 7th Apple Countdown
The InformaTIon scored the above companies mainly from three indicators: technology, engineering and business model. Among them, Waymo is the only company that has a technical score of 5 points, and a large distance from the second place, which means that Waymo has a technical lead. At the same time, only Waymo and Uber scored 4 points or more in the project.
The InformaTIon analysis believes that technology companies such as Waymo and Uber have established significant advantages in key areas of driverlessness, leaving traditional car companies behind.
Uber is the only company that has achieved full marks on the business model. The Chinese version of Uber's Didi Travel also scored 4.5 points.
At present, the industry generally believes that the combination of driverless and shared travel in the future will bring greater business opportunities. Most of the companies on the list have a layout for sharing travel or bus services. For example, GM owns Maven and acquires a large stake in Lyft, BMW owns DriveNow, and Daimler owns Car2Go.
The list also evaluated two Chinese companies – Baidu ranked seventh with a total score of 8.5, and averaged among the three indicators. At present, autonomous driving has been promoted to a higher strategic position within Baidu. The company launched the "Apollo Program" in April this year, which will open the autopilot software to the entire industry.
Another Chinese company is Didi Chuan, but it ranks the last among 17 companies, especially with a technically unique zero. However, this may be related to the late start of the drone in deploying driverless, and the company is already increasing its investment in the field. It is reported that Didi traveler currently hired Uber's famous safety engineer Charlie Miller, and dug up Waymo's engineers specializing in object detection.
Surprisingly, the tech giant Apple has become the "trailer tail" on the list - a total score of 5.5, tied for 15 with Fiat Chrysler. Apple was one of the early unmanned technology companies. Two years ago, the company launched the dedicated unmanned vehicle project "Project TItan" and established a special department. But unfortunately, the project has progressed quite slowly and the specialized automotive sector was abolished last year. Foreign media sources said that Apple may abandon plans to build cars independently and cooperate with mature car manufacturers to develop.
Taken together, the findings of The Information show that in the higher-level autopilot field, traditional automakers do not have an advantage, which is in line with the previous ranking of another institution.
In April of this year, the global autopilot global strength ranking released by market research institute Navigant Research showed that the traditional automakers had an overwhelming advantage. Ford ranked first, followed by GM, and the other three of the top five were respectively. For the Renault Nissan Alliance, Daimler and Volkswagen. Google, the star company in the eyes of the public, is only ranked seventh, while Uber and Baidu, both of which are tech companies, have not even entered the top ten.
Navigant Research's report explores the strategy and implementation of 18 leading companies developing autonomous driving systems to objectively assess the relative strengths and weaknesses of the global autopilot system market. It contains 10 criteria for judging: vision; market strategy; partner; production strategy; technology; sales; marketing and distribution; product capability; product quality and reliability; product introduction;
The report divides autonomous driving companies into four categories: lead, competitor, challenger and follower. The three leading companies – Ford, GM, Daimler – are in fact the giants of the traditional automotive industry. Google’s parent company Waymo and Tesla were classified as competitors, while Baidu and Uber were placed among the third-tier challengers.
The report also pointed out that competition in the field of autonomous driving is a protracted arms race. With the development of technology and the release of future strategies of these companies, various factors will affect the rankings. It is difficult to have a permanent victory in this field. The company will present a dynamic ranking.
How much is American driverless technology ahead of China?Autopilot is a matter of time, and now it seems that the popularity of autonomous driving is faster than expected.
Once autopilot is widely used, many functions will become a reality. For example, vehicles can be grouped and run like a moving car to achieve synchronous high-speed traffic. Communication and interaction between vehicles can achieve efficient traffic efficiency and extremely low The accident rate does not even require traffic lights and various complicated traffic sign marking systems. The overall traffic efficiency and safety will be greatly improved and revolutionary.
It won't take long for someone to say, "It's so dangerous to drive a car. It used to be so scary. It's unimaginable." Of course, the term "female driver" may also have to withdraw from the historical arena.
Many friends are worried about transportation infrastructure and legal responsibilities. I don't think this is too much of a concern. The process of automatic driving is also a process of continuous improvement of related transportation technology facilities and laws.
Today, we mainly analyze the technical aspects of autonomous driving. Other related road facilities and legal responsibilities are not discussed here.
First, let me briefly talk about the principle of autonomous driving.
All control systems are composed of sensors, controllers and actuators. From this point of view, the principle of autonomous driving is the same as that of manual driving. We use the eyes to observe the road conditions, while the automatic driving uses laser radar and ultrasonic waves. Radar, camera, GPS and other sensors to observe the road conditions to determine the location. We use the brain to make judgments. Autopilot is of course judged by using a computer as a controller. Then we control the steering wheel, acceleration and braking of the vehicle through the hands and feet. The automatic driving also controls the vehicle directly according to the output of the computer.
It looks very simple, but if you want to analyze it in depth, especially if the car is so dangerous, it is very complicated to ensure the reliability of autonomous driving.
On the sensor side, Lidar and ultrasonic radar are of course used for distance measurement. Ideally, the vehicle can detect all obstacles around and calculate the distance of these obstacles, but in reality, these obstacles are likely to be misjudged. For example, a plastic bag flying in the wind will be judged as an obstacle, and even a raindrop may be considered an obstacle. Needless to say, the monocular or multi-camera uses computer vision to allow autonomous vehicles to recognize traffic lights, traffic signs, lane lines, close-range low-speed obstacles, etc. in real time, plus communication with road infrastructure and cloud databases. Achieve many functions. However, the death toll from Tesla's automatic driving this year is largely related to the camera. In the case of backlight and large light ratio, the resolution of the camera is reduced, and of course there is low illumination. This is limited by the current image sensor technology. Friends who know photography will know that machine vision is a complicated thing. At the same time, the camera may be affected by various unfavorable factors such as dust and glare, and all of them still have many problems in terms of reliability. Of course, in addition to these sensors that determine road conditions, the car also has various sensors such as speed, acceleration, and angle of rotation.
In the control center, receiving so many sensors information, performing analysis and processing, obtaining control strategies, issuing control commands, and this process must be real-time, so the control center must have high-performance real-time computing capabilities. As far as hardware is concerned, the process of autonomous driving can not be allowed to crash, so the general hardware must use real-time computing that satisfies high reliability and high performance. As far as software is concerned, the algorithm is too important. The software algorithm of autopilot must implement path planning, obstacle avoidance, acceleration control, attitude control and so on. However, there is no control method that can achieve perfection. Driving, many times manufacturers use a variety of methods to deal with, such as fuzzy control with genetic algorithms, deep learning and so on. As the answering person who also wrote a lot of control software, I feel that this sentence is really right: "99% of the cases can be dealt with 1% of the code, and the remaining 1% of the situation needs 99%. The code goes to process."
In terms of actuators, once the control instructions are obtained and the car is executed, this part is still quite good at present, especially for electric vehicles. The extremely high efficiency of the motor, the excellent speed regulation performance and the wide speed regulation interval make it easier for the electric vehicle to achieve automatic driving. The ordinary car has developed the engine automatic control system, automatic gearbox and electronic brake system after years of development. The control has also matured, and systems such as ESP, TCS, cruise control, and adaptive cruise have also been widely used, and it is not troublesome to interface with the automatic control system.
Let's talk about the situation of unmanned research in China and the progress of research in the United States, and then look at the gap.
domesticAs early as 1992, the National Defense Science and Technology University successfully developed China's first truly driverless car. In 2007, the company jointly developed the Red Flag driverless car with FAW. The car mainly uses CCD image sensor and laser radar as sensors to realize unmanned driving on the highway. In 2011, HQ3, developed by the National Defense Science and Technology University, realized a road test of 286 kilometers from Changsha to Wuhan.
Jilin University has developed the JLUIV-1 driverless car, using fuzzy control plus genetic algorithm correction.
In 2005, Shanghai Jiaotong University cooperated with the European Union on the Cyber ​​C3 project to study regional traffic intelligent vehicles for urban environments.
In 2012, the Military Traffic Academy's "Lion 3" equipped with 5 radars, 3 CCD image sensors, and 1 GPS, traveled 114 km in an unmanned state with a top speed of 105 km/h.
In 2016, Beijing Institute of Technology designed an unmanned racing car that uses a binocular camera for local route planning such as pedestrian detection and obstacle avoidance. Interestingly, this car is only 3 seconds faster than 100 kilometers.
In 2013, Baidu also started to engage in self-driving cars. In early December 2015, Baidu driverless cars carried out full-automatic driving test in Beijing, achieving complex driving actions such as multiple car deceleration, lane change, overtaking, up and down ramps, and turning around. The switchover of different road scenes from high speed to exit high speed is completed, and the top speed is up to 100km/h. The picture below is Baidu's driverless car.
On September 1 this year, Baidu announced that it had obtained the 15th driverless test license in California. Baidu's investment in unmanned driving is huge, including the establishment of Silicon Valley R&D center, and Ford's investment in laser radar manufacturer Velodyne (providing laser radar for Google). Baidu also announced its "three-year commercial" and "five-year mass production". "The goal.
The research on the unmanned technology that has just started in China has become the focus of many high-end talents. Jiang Yan, the person in charge of the unmanned driving vehicle from Beijing Institute of Technology, and the former Intel China Research Institute President Wu Gansha It is a company specializing in autonomous driving research. It has deep technical accumulation in binocular vision, provides low-cost pure vision autopilot solutions, and has made considerable progress in the autonomous environment of restrictive environment. Of course, there are still many similar small companies that are conducting research on autonomous driving technologies from different levels.
The above-mentioned technologies are basically based on CCD image sensors, radar ranging and other technologies. Under simple high-speed road conditions, almost all of them can achieve good autonomous driving performance. However, the urban road conditions are still very much. Domestic autopilot is still in its infancy, and it is still in the prototype verification stage of R&D. The gap with the United States is not small (not only automatic driving, but also in many aspects). Baidu is regarded as the domestic leader in this respect, and cooperates with NVIDIA, and also cooperates with the government to promote autonomous driving technology. But from the distance of the road test, Baidu is also far worse than Google. The emergence of a professional company like é© ç§‘æŠ€ technology has also added a strong force to domestic autopilot research.
United StatesOf course, the most representative of the most cattle are Google and Tesla.
Google has always been the world's top software, although Microsoft and Apple may not be convinced, but I think so.
So Google's powerful strengths are software and algorithms. Here is Google's self-driving car.
Google's self-driving cars have GPS, cameras, radar and laser sensors that can capture information from the surrounding environment in a 360-degree view. Since 2009, Google's self-driving cars have traveled more than 1.2 million miles in autonomous mode. The software already knows a lot about how to deal with different situations. The picture below is the world in the eyes of Google's self-driving car.
It can be seen from the figure that the various sensors of the body can detect objects as far as two football fields, including people, vehicles, buildings, birds, bicycles, etc. This car can see other vehicles. These vehicles are shown in purple in the figure, the cyclists will be marked in red, and the corners in the upper left corner will be indicated by orange cones. It can even recognize traffic police gestures, which is very remarkable and is a reflection of Google's powerful software algorithm capabilities. Although Google's self-driving cars can predict a lot of things based on the collected data, many times more powerful than the domestic ones, there will still be situations that have never happened before. On one occasion, a self-driving car under test was driving in the Mountain View area. A woman sitting in an electric wheelchair was chasing a duck on the road, but the car could only continuously test and slow down. To escape this woman.
Tesla is already mass-produced and put into mass use while others are still researching and experimenting.
As pure electric, Tesla has an advantage in automatic driving convenience, and Tesla is particularly attached to autonomous driving technology. The current Tesla model on the road, Autopilot assisted driving uses 12 ultrasonic sensors around the body to identify the surrounding environment, a front camera to identify the front object, a front radar to identify the front object, and It is a high-precision satellite map that has been accumulated for a long time. This achieves the "partial autopilot" function in the table below.
Tesla is very excited to announce on October 20 this year: Starting today, all the Tesla models produced at the factory, including the Model 3--, will be equipped with hardware with fully automatic driving functions compared to manual driving. New hardware will fundamentally improve the safety of vehicle driving. The system will consist of 8 cameras covering a 360-degree viewing range and monitoring distances up to 250 meters from the surrounding environment. In addition, the vehicle's 12 ultrasonic sensors complete the vision system, detecting and sensing hard and soft objects at twice the distance of the previous generation system. The enhanced front-end radar provides richer data around the redundant wavelengths, and radar waves can travel through heavy rain, fog, dust, and even vehicles ahead. In order to make better use of this data, the vehicle is equipped with a processor 40 times faster than the previous generation, running Tesla's vision system, sonar and radar system software based on deep neural network. In summary, the system provides a world image that the driver can't reach with the eyes, and detects each direction of the vehicle through the band synchronization, which is far beyond the human senses.
Tesla covered the sensor with the entire car. The eight cameras provide 360-degree visual monitoring, with the ability to monitor objects up to 250 meters, and 12 ultrasonic sensors covering twice the range of previous Autopilot systems. An enhanced version of the radar is used to detect vehicles in front of the rain and snow. These hardware enable Tesla's autonomous driving to be "highly automated".
In addition, the models running on Tesla Road use their various sensors to contribute to Tesla's high-precision satellite maps. This is not comparable to Google. After all, Google is running around with several test cars. Going, and Tesla's mass production models are to some extent test cars. At present, Tesla's Autopilot assisted driving has reached 222 million miles (about 357 million kilometers).
So overall, whether it is Google or Tesla, its autonomous driving technology is more advanced than domestic, and it will take time to catch up.
Sometimes, the gap is that, when you look at it, you can catch up with it, and then you can work hard. When you just want to catch up, the goal is also accelerated, and the gap is even bigger.
Display Port And Mini Display Port
Display Port And Mini Display Port
Display Port ,Mini Display Port
ShenZhen Antenk Electronics Co,Ltd , https://www.antenksocket.com