When it comes to hoarseness, many people's first reaction is underwater navigation system. Indeed, sonar technology should be generated by underwater target early warning detection. It is a kind of propagation characteristics of sound waves under water, through electroacoustic conversion and information processing. An electronic device that performs underwater detection and communication tasks. However, sonar is not the exclusive object of underwater detection. In the terrestrial environment, people have always wanted to apply the sonar, but so far, they have only slowly moved forward in the initial stage of emulating bats.
A few days ago, Tel Aviv University said that the researchers developed a completely autonomous terrain robot called Robat, which can make sounds like bats and analyze echoes to identify, draw and avoid outdoor obstacles. Robat is the first fully autonomous, biologically bat-like walking robot. The advent of this terrain robot undoubtedly adds a new coordinate to the sonar technology circle.
"Amphibious" technology: terrestrial applications for underwater sonar
This is a very simple physical effect. Usually, the speed of sound propagation in a solid is greater than the speed of propagation in a liquid, and the speed of propagation in a liquid is greater than the speed of propagation in a gas. The working carrier of the sonar is the sound wave. The sound wave travels faster in the water, the energy decays slowly, and the detection position does not change greatly, which is relatively accurate.
However, in view of the advantages of sonar technology ranging distance and anti-interference, people have been trying to apply the sonar system to land robots to generate machine vision concepts and recognize the shape of surrounding objects. But the effect of such an attempt is not good. Technically, it is still difficult to overcome the problem that high-frequency sounds are attenuated by the influence of distance and reflection during propagation.
People are eager to develop the "skull" of the sonar into a "frog", and the "animal" thing is naturally solved by animals. Through observations of bat bioacoustic sputum, research shows that some of the biosonic organ of the bat's head (such as the auricle, tragus, etc.) will propagate in the process of ultrasonic signal propagation during the ultrasonic propagation. It affects the sound field of the bat's launch.
This also gives people a way to think. If you can extract the trajectory of the bat's auricle movement, perform three-dimensional reconstruction, and apply these "organs" to the robot, can you solve some problems?
If you use the ultrasonic speaker or transmitter to create the pronunciation "mouth", use two ultrasonic receivers to create the radio "Auricle" or "Tin Screen", the pronunciation "Mouth" sounds around at a steady frequency, and uses "Auricle" or The "ear screen" adjusts the signal, affecting the distribution of sound pressure and the distribution of sound waves in the external environment, so that the robots active in the outdoor environment can obtain real-time mapping images around.
The best example is to solve the pain point of agricultural production forecast. The sonar system evaluates the growth environment of crops through the acoustic characteristics of different objects in the crop garden plot, and uses intelligent algorithms to analyze and calculate the sound wave results. Quantitative predictions about production. Researchers from Israel have recently launched a new project to predict the production of crops such as fruits through the sonar robot "AGRYbot".
Radar, lidar, sonar... The spring of automatic driving is coming?
Bats are a group of animals. Usually, many species and a large number of bats live in caves or woods. In the case of southern China, there may be tens of thousands of bats in a cave. . In a complex environment, there is no signal confusion between the bats. These bats will continue to go out for food in the evening.
The beneficiaries of this feature are undoubtedly the hottest driverless at the moment.
In the technical unit of unmanned vehicles, the detection, prediction and collision of obstacles in outdoor complex environments have always been difficult to study. At this stage, the single method of identifying contour features has low accuracy and speed in the process of dynamic obstacle detection and tracking, identifying the types of dynamic obstacles, accurately detecting and tracking dynamic obstacles and predicting dynamic obstacles. The performance of the trajectory is not satisfactory.
Take the laser radar technology most commonly used in unmanned driving as an example. Lidar is greatly affected by the weather during work. The laser is less attenuated in clear weather and has a long distance. In heavy rain, heavy smoke, dense fog and other weather, the attenuation will increase sharply and the propagation distance will be affected. Great impact. For example, a CO2 laser with a working wavelength of 10.6 μm is better for atmospheric transmission in all lasers, and the attenuation in bad weather is six times that of sunny days. Moreover, the atmospheric circulation will also cause the laser beam to be distorted and shaken, which will directly affect the measurement accuracy of the laser radar.
Even radars with farther distances and higher prices face many problems. As early as 2016, the radar was applied by Tesla Autopilot 8.0, but Tesla also publicly stated that although the radar as the main detection means of Autopilot helps prevent similar traffic accidents, it also has the ability to detect people's clarity. The degree is not as good as the camera, the recognition of wood and plastic objects is not in place, and the reflection of the metal object causes problems such as receiving information errors.
The sonar has certain advantages in the recognition of different objects. The passive sonar in the sonar can also directly accept the noise of the external environment or other mechanical work, and the concealment and exclusivity are better. However, ultrasonic waves emitted by sonar sensors often have energy attenuation problems in long-distance detection and are susceptible to specular reflection.
Each of them has its own advantages and disadvantages, and neither of them can achieve the best results. If radar or lidar is combined with sonar, it may improve the vehicle's ability to detect and judge obstacles, thus ensuring the accuracy of vehicle routing and selection, and effectively improving the self-adaptation ability of unmanned driving to complex environments. And adjustment ability.
For example, when the lidar is combined with the sonar, the vehicle travels forward, and the sonar sensor and the laser sensor work simultaneously. When the sonar sensor detects an obstacle in front, it should be combined with the detection result of the laser sensor. If the results of the two tests meet the characteristics of the obstacle, the coordinates of the obstacle are marked in the field of view of the vehicle, and multiple times are detected. When the information is consistent, the distance and orientation information of the obstacle is determined and recorded, and then the vehicle selects the correct path to continue according to the path selection strategy.