Automatic driving and trouble, the study of MIT or will change the fate of laser radar

In recent years, automatic driving technology has swept the globe, and is considered to be the next innovation in the field of science and technology. In addition to the large-scale layout of traditional car manufacturers, technology companies have played an important role in promoting Google, Uber, Baidu and NVIDIA. With the continuous progress of technology and massive entry of capital, automatic driving has begun to come true. However, this process is not smooth sailing, and even suffered a fatal blow.
Uber self driving accident. Lidar is "back pot man".
At 10 p.m. Eastern time in March 19th, a Uber unmanned vehicle killed a female pedestrian in tanpe, Arizona, which was the first known case of a pilotless car causing a pedestrian death. After the accident, Lior Ron, co-founder of Uber Uber Freight, left the company and halted the automatic driving test.
After many days of continuous fermentation, this event has cast a shadow on the development of automatic driving technology. After all, it has involved pedestrian safety. To this end, TOYOTA has suspended automatic driving test projects in the United States, and Jen-hsun Huang, the co - founder and chief executive of the NVIDIA, said that all autopilot tests should be suspended.
Up to now, the latest news shows that Uber has reduced the number of laser radars from unmanned test vehicles from 5 to 1 on the top of the car, causing more blind spots and unable to fully detect pedestrians. This is considered the main cause of the accident. Velodyne, a laser radar supplier, also believes that radar must be installed on both sides of the vehicle to avoid hitting pedestrians. Though the motivation of Velodyne's statement is mostly to get rid of the need as far as possible, the importance of lidar in automatic driving is self-evident.
MIT's research will change the fate of the lidar
In addition to the camera and other sensors, in addition to generating a three-dimensional position model, the lidar has a farther detection distance, higher measurement precision, and more sensitive response speed, and is not affected by environmental light. Moreover, the data of lidar feedback is easier to read and analyze by computer, so it becomes one of the core technologies of automatic driving. However, the laser radar is also easily affected by the weather, especially in the rain and fog weather will cause the laser scattering, thus affecting the precision of the laser radar, the detection distance is inaccurate.
In order to solve this problem, MIT developed a new imaging system. In fog, laser can accurately calculate the distance. As we all know, the laser radar is mainly to calculate the distance from the front object by the short pulse laser emitted and the reflection time reflected back from the front object. Due to the scattering phenomenon of heavy fog, the time of reflecting the laser beam is often different from that of the clear day, which leads to the inaccuracy of lidar detection. MIT researchers found that no matter how thick fog may be, scattering laser reflection time always followed a very specific distribution pattern. When the camera calculates the number of photons returned to the sensor at a frequency of 1/1000000000000 seconds and draws the results into graphs, the system can use a specific mathematical filter to find the peak of the data to get the exact distance of the actual objects hidden in the fog.
As a result, we will be able to set up a model by testing and modify the lidar detection to improve its detection accuracy in the dense fog weather and clear the obstacles for the application of laser radar in the field of autopilot.
Automatic driving technology development is the great scourges
Uber unmanned vehicle caused a pedestrian death accident, different reactions, and even some people began to oppose autopilot technology, that the technology will become a "pedestrian killer". In this regard, the attitude of the industry is more pragmatic.
"Automatic driving must be a safer way to travel," Robin Li, chairman of Baidu, said in the "China high development forum". "A car accident in China will kill 500 people every day, but no one has reported that a Uber auto driver killed a person and knew all over the world." I believe that as time goes on, people will realize more and more that automatic driving is safer than human driving.
Li Shufu also said in the "2018 intelligent automobile International Symposium" that "advanced autopilot from mature technology to real people's life, really change our daily trip, there is a long way to go, and doomed not to be one sail smooth. No matter how the development of automotive technology, safety is a problem that can not be ignored at any time.
At the same time, autopilot processing company Mobileye calls for the establishment of a "verifiable security" standard model to verify the effectiveness and safety of all autopilot technologies, especially Uber, a new technology company that enters the field of autopilot.
After the accident, Jen-hsun Huang also said: "such an event, you need to learn new information, need to suspend and learn from it. I think everyone in the industry should do this. First stop and learn from it.
The technology of automatic driving is not a flood. It will be like other technologies, and it will better serve the life of mankind after being perfected. This unfortunate accident, as you may say, is that Uber has gone too fast.