Where is the unmanned ground platform "cow"?
With the development of science and technology, the comprehensive development of artificial intelligence innovation applications has brought unprecedented impacts on the intelligence of all aspects of human society. With its excellent technical performance, the unmanned ground platform has become the commanding height of the intelligent field that all major countries in the world are competing for, and has also become the development direction of automation and intelligence of military equipment in various countries.
The unmanned ground platform combines many “skills”. Among the powerful capabilities such as sensing technology, map reconstruction technology, navigation technology, learning and adaptive technology, the most prominent is the perception of the surrounding environment and map reconstruction technology. With these "magic weapons", the unmanned ground platform can become "clear eye" and "shunfeng ear", providing strong technical support for command and decision.
Let's take a look at the environment-aware technology, which is part of the ability to recognize roads on unmanned ground platforms. Through the sensor, the unmanned ground platform can collect the surrounding and its own information, and send it to the processor in real time to form a cognitive model for the surrounding environment. It can be said that environmental awareness is the data foundation of all technologies of the unmanned ground platform. Let's talk about the map reconstruction technology, which can identify, process and analyze the real-time environment-aware feedback information, and constantly update and improve the existing map information to effectively cope with unexpected situations.
Despite such a powerful "housekeeping skills", traditional single sensors are difficult to guarantee the reliability of the unmanned ground platform environment under complex conditions. For example, high-precision GPS, once in some cases, the satellite signal is interrupted or severely interfered, making the satellite signal very weak, the environment perception is facing great difficulties; the vision-based environment-aware technology encounters strong light or low light. When the illuminance changes, the perceptual algorithm will be invalid. Based on the laser-based environment-aware technology, due to the limited detection distance, sometimes there is a lack of sufficient target attribute information, resulting in the unmanned ground platform becoming the "scorpion" and the "scorpion".
To solve these problems, the researchers installed a variety of sensors on the unmanned ground platform, enabling it to have multi-sensor data fusion sensing capabilities, merging each picture through countless pairs of "eyes" and countless pairs of "ears". Split the main elements in the scene and clearly distinguish them one by one. Researchers at Michigan State University have used this method of installing multiple sensors to effectively sense all road conditions on campus through the integration of sensors, making the unmanned ground platform a veritable path to the campus. . The research team of Nanyang Technological University in Singapore focuses on the scene semantic segmentation technology of laser radar and visual data fusion. Through the fusion method, the main elements in the scene can be well segmented and their semantic information, including the target vehicle, Pedestrians, ground, grass and trees. In addition, companies working on driverless car research, such as Google and Ford, have applied positioning and map construction technology to achieve high-precision map construction and centimeter-level precise positioning, but their application environment is limited to urban structured roads.
Although the current ground-based unmanned platform has superior capabilities, it still has great limitations in the use of complex environments in the field. Especially under special conditions such as war or major disasters, the environmental perception of various complex obstacles is far from reaching the limit. The sensitivity of people's satisfaction. This also poses new challenges for further optimizing multi-sensor fusion sensing technology for unmanned ground platforms.
The unmanned ground platform combines many “skills”. Among the powerful capabilities such as sensing technology, map reconstruction technology, navigation technology, learning and adaptive technology, the most prominent is the perception of the surrounding environment and map reconstruction technology. With these "magic weapons", the unmanned ground platform can become "clear eye" and "shunfeng ear", providing strong technical support for command and decision.
Let's take a look at the environment-aware technology, which is part of the ability to recognize roads on unmanned ground platforms. Through the sensor, the unmanned ground platform can collect the surrounding and its own information, and send it to the processor in real time to form a cognitive model for the surrounding environment. It can be said that environmental awareness is the data foundation of all technologies of the unmanned ground platform. Let's talk about the map reconstruction technology, which can identify, process and analyze the real-time environment-aware feedback information, and constantly update and improve the existing map information to effectively cope with unexpected situations.
Despite such a powerful "housekeeping skills", traditional single sensors are difficult to guarantee the reliability of the unmanned ground platform environment under complex conditions. For example, high-precision GPS, once in some cases, the satellite signal is interrupted or severely interfered, making the satellite signal very weak, the environment perception is facing great difficulties; the vision-based environment-aware technology encounters strong light or low light. When the illuminance changes, the perceptual algorithm will be invalid. Based on the laser-based environment-aware technology, due to the limited detection distance, sometimes there is a lack of sufficient target attribute information, resulting in the unmanned ground platform becoming the "scorpion" and the "scorpion".
To solve these problems, the researchers installed a variety of sensors on the unmanned ground platform, enabling it to have multi-sensor data fusion sensing capabilities, merging each picture through countless pairs of "eyes" and countless pairs of "ears". Split the main elements in the scene and clearly distinguish them one by one. Researchers at Michigan State University have used this method of installing multiple sensors to effectively sense all road conditions on campus through the integration of sensors, making the unmanned ground platform a veritable path to the campus. . The research team of Nanyang Technological University in Singapore focuses on the scene semantic segmentation technology of laser radar and visual data fusion. Through the fusion method, the main elements in the scene can be well segmented and their semantic information, including the target vehicle, Pedestrians, ground, grass and trees. In addition, companies working on driverless car research, such as Google and Ford, have applied positioning and map construction technology to achieve high-precision map construction and centimeter-level precise positioning, but their application environment is limited to urban structured roads.
Although the current ground-based unmanned platform has superior capabilities, it still has great limitations in the use of complex environments in the field. Especially under special conditions such as war or major disasters, the environmental perception of various complex obstacles is far from reaching the limit. The sensitivity of people's satisfaction. This also poses new challenges for further optimizing multi-sensor fusion sensing technology for unmanned ground platforms.
No comments