TU-TESL

• 22

According to foreign media reports, vehicle navigation has been improved all over the world, but before the popularity of autopilot cars, we need to solve a series of problems first. One of them is bad weather conditions, which may affect the sensor technology used in the current automatic driving vehicle. This sensor technology can determine the location of vehicles and the location of vehicles in other animal moving objects. The technology developed by Geophysical Survey Systems (GSSI), called TerraVision, uses GPR to improve the navigation of autopilot. Dr. David CIST, vice president of engineering of GSSI company, said that unlike ordinary ar sensors, GPR has the advantage of not being affected by weather conditions. "Rain, fog and mud, or dirty windows will disturb the signal and the sensor will fail. Snow will cover the road, covering the lane paint, road signs and other signs used for navigation. Although, in addition to the weather impact, if there is no building or tree shelter, GPS navigation effect is also very good. However, night conditions, especially rainy nights, can pose challenges to navigation. In addition, seasonal changes will also have a great impact on navigation. " GSSI's lgpr Technology (localization ground penetrating radar) creates a digital fingerprint underground by sending radio waves to the ground, which is used to locate vehicles equipped with terravision. Because the electric wave can penetrate three meters underground, and will reflect rocks, roots, soil and pipes. The reflected signal will be used to build a 3D map of the underground situation. Splicing these images together creates a complete 3D fingerprint that can be used by any vehicle equipped with lgpr technology to determine its location. In addition, the creation of a complete map of the ground and underground enhances the possibility of locating and navigating an autonomous vehicle under any conditions. Almost all autopilot car navigation relies on 3D maps created by cameras and lidar. Then, the cameras equipped with the autopilot car and the lidar sensor can be used to locate the map with road markings, street signs, buildings and other known locations. Terravision is the same, except that the generated 3D map is about the underground. With such a map, any vehicle equipped with terravision knows its geographical location. When the vehicle equipped with terravision can find the corresponding data or fingerprint on the reference map, it can know its location. When the vehicle moves, the next data may find the corresponding data in the map again. For each match, the direction and speed of the vehicle can be calculated to track the vehicle. Like other maps, terravision reference maps need to be geolocated in some way, either through an accurate GPS or integrated into other ground maps to form a powerful navigation system to reduce failure. Years of experiments at MIT have demonstrated the potential of lgpr technology. Now, GSSI improves its performance with faster hardware, smarter software, and smaller mechanical designs. GSSI redesigned the RF switching, reduced the power consumption by four times, reduced the radio wave emission by more than 100 times, significantly reduced the size, made it more weather-free, and improved its performance. In late June, terravision was successfully integrated into the L2 autonomous driving test vehicle in a closed-loop field test in Devens, Massachusetts. L2 autopilot requires the driver to sit in the driver's seat to control the vehicle at any time. However, the L2 system can control steering and acceleration. Considering the uncertainty of FCC (Federal Communications Commission) restrictions in the United States, many companies around the world (including two large Japanese and German companies dedicated to autopilot navigation) require GSSI to develop and test lgpr, and GSSI initially decided to focus on overseas. At the end of this year, GSSI's performance test of lgpr technology in Germany will determine whether the technology is effective. With years of professional knowledge and data, GSSI can prove that GPR map is still absolutely reliable for more than ten years, and GSSI's geophysics experience can also ensure that lgpr map is stable and reliable. The LGPR test shows that the lane location accuracy is about 4cm when driving at the speed of the highway, which is equivalent to the accuracy of other automatic driving vehicle navigation sensors. For example, GPS navigation accuracy is 30cm, except for cities, forests and tunnels. The lidar, radar, camera and other sensors can scan the characteristics of the ground and the surrounding environment with an accuracy of about 10cm, but such navigation is easily damaged by rain, snow, dust, fog and even leaves. Adding LGPR technology to automatic driving vehicle can compensate for life threatening sensor failure. If the international LGPR field test can confirm the results of MIT and GSSI, it will be a great victory for the auto driving industry, because the factors that affect the acceptance of the automatic driving vehicle is that the self driving vehicle needs to be positioned without fault under all conditions. All sensors have limitations, and terravision is no exception. However, if the automatic driving vehicle reaches L3, the sensor equipped with it must be able to make up for each other's deficiencies under all conditions. (all pictures in the article are from spar3d. Com)