MIT develops new technologies: driverless cars can navigate without maps.

category:Internet
 MIT develops new technologies: driverless cars can navigate without maps.


Photo: CSAIL, a laboratory under Massachusetts Institute of Technology, is equipped with a new system that can run on rural roads for the TOYOTA Prius. The framework is called MapLite, which combines the global positioning system (GPS), uses the most basic topographic maps from OpenStreetMap, and laser radar and IMU sensors used to monitor road conditions. CSAIL graduate student Teddy Ortega (TeddyOrt) said: this no map model has been tried before, because it is often difficult to have the accuracy and reliability of a detailed map. A system like this can navigate on a vehicle sensor, which shows the potential of an unmanned vehicle, and it can cope with the road beyond the map made by a technology company. Researchers have drafted a report to describe the system, which will be submitted to the International Conference on Robotics and automation (ICRA), held in Brisbane, Australia later this month. To test the system, the researchers equipped lidar, sensors and MapLite for the unmanned TOYOTA Prius. This Prius can see 30 meters ahead of the road, successfully in a number of unpaved roads in Massachusetts. The researchers said in the report: it means that it can drive the car at more than 100 kilometers per hour, and if the method can be parallelized and realized on the GPU, the speed of the car may be faster. Ort explains that their systems are also different from other no map patterns, which use machine learning to train the system. They use data from a group of roads, then test their patterns and apply them to other roads. Instead, MapLite tries to develop a model of what a pilotless car might encounter, and then tell its behavior, which is not very different from the way human drivers drive in a familiar scene. Ultimately, we want to be able to ask more questions about cars, such as how many roads are merging at this intersection? using modeling techniques, we can better understand the reason if the system does not work or have an accident. Although a system like MapLite can open up more directions for the development of driverless cars, it still has a long way to go. The researchers said that their system could not explain the changes in altitude, so they could not cope with mountain roads now. I think the future of unmanned vehicles will always use 3D maps in urban areas, Ortega said. And when we drive on a remote country road, these vehicles need to be like human beings on a strange road. We want our efforts to take a big step in this direction. (small) source: NetEase science and technology report editor: Wang Fengzhi _NT2541 Ultimately, we want to be able to ask more questions about cars, such as how many roads are merging at this intersection? using modeling techniques, we can better understand the reason if the system does not work or have an accident. Although a system like MapLite can open up more directions for the development of driverless cars, it still has a long way to go. The researchers said that their system could not explain the changes in altitude, so they could not cope with mountain roads now. I think the future of unmanned vehicles will always use 3D maps in urban areas, Ortega said. And when we drive on a remote country road, these vehicles need to be like human beings on a strange road. We want our efforts to take a big step in this direction. (small)