This article is from autonavi’S wechat official account (AMap_tech).
The background,
High precision map, high precision collection car, map and travel field students often hang in the mouth of some common words. But, those of you outside the circle might ask, what exactly is an elite?
High precision refers to high precision positioning, and high precision map refers to maps with rich geographic information data and high precision coordinates. Of course, the high-precision acquisition vehicle is a special operation vehicle for collecting and making high-precision map data.
Some curious friends will break out of the casserole and ask, how is the high precision achieved? How to call high precision?
In fact, the current high precision standard is not very certain, but basically think that the accuracy of the centimeter level above can be regarded as high precision. The realization of high precision mainly depends on all kinds of sensors, among which the most important is the high-precision positioning and orientation system, including satellite positioning and inertial navigation two parts.
This paper mainly introduces some of this aspect from the point of view of hardware, as well as its application in practical work.
2. Explanation of terminology
Positioning and orientation system: Position and Orientation System (POS) refers to a high-precision Position and attitude measurement System combined with inertial navigation and GNSS satellite navigation. The satellite receiver installed on the carrier is used to accurately determine the spatial Position, and the inertial measurement device is used to determine the instant sensor attitude. By combining them with an accurate clock, the speed, attitude and position of the carrier can be obtained through calculation.
Inertial navigation system (INERTIAL navigation system for short) is a navigation parameter calculation system based on gyroscope and accelerometer. According to the output of gyroscope/accelerometer, a navigation coordinate system is established to calculate the speed and position of the carrier in the navigation coordinate system.
IMU: Inertial Measurement Unit, a device that measures the attitude Angle (or angular velocity) and acceleration of a triaxial object. The IMU is part of the inertial navigation system.
* * GNSS: ** Global Navigation Satellite System, a general term for all Satellite Navigation systems, including Global, regional and enhanced ones, such as GPS of the United States, Glonass of Russia, Galileo of Europe, And Beidou of China; And related augmentation systems, such as WAAS (Wide Area Augmentation System) in the United States, EGNOS (European Stationary Navigation Overlap System) in Europe and MSAS (Multi-function Transport Satellite Augmentation System) in Japan, as well as other satellite navigation systems under construction and to be built in the future.
Note: We are used to refer to the positioning part as inertial navigation for short. In fact, it not only includes inertial navigation equipment, but also refers to the positioning and orientation system including the whole set of hardware and software.
Three, what does the high-precision positioning and orientation system include
The high-precision positioning and orientation system on the acquisition vehicle is generally composed of the following parts:
Composition of orientation system
The algorithm process
The whole system is composed of hardware and supporting software and algorithm. Since there are many researches on the algorithm of combined solution in the industry, all kinds of methods and ideas are blooming, and there are related students in Autonavi to do the development work in this aspect. Therefore, this paper only focuses on the introduction of hardware.
Low precision, high precision positioning system composition is similar, the difference is only the sensor (IMU, GNSS) accuracy level is different.
Four, the function and role of each sensor
1.GNSS
Similar to GPS often said before, but with the progress of science and technology, our Country’s Beidou System BD can match GPS in accuracy and reliability, playing an important role in practical application.
From modules with tens of yuan to high-precision GNSS boards with tens of thousands of yuan, the positioning principle is basically the same, which is to measure the distance between the known position of the satellite and the user receiver, and then integrate the data of multiple satellites to know the specific position of the receiver. The position of a satellite can be found in a satellite ephemeris based on the time recorded by the onboard clock.
However, in terms of SNR, frequency band, number of constellations, number of channels, and signal acquisition and tracking capabilities, high-precision GNSS board cards are significantly superior to ordinary GPS. For example, mobile phones generally only support THE C/A code of GPS L1 band, but professional board cards basically support L1/L2/L5 multi-band and multi-channel.
In addition to being significantly better than ordinary modules in real-time positioning accuracy, professional GNSS boards can also be post-processed with accuracy up to millimeter level, which is where the value of professional receiver boards lies.
Major satellite receiver manufacturers in the industry include foreign Trimble, Novatel, LeiCa LeiCa, Topcom, etc., as well as Domestic BeidouxingTong, Huazao, Zhonghaida, Sinan, etc.
The main function of GNSS in the system is to obtain the absolute coordinates of the current position. The advantage is that there is no position error accumulation, and the disadvantage is that the update frequency is low, generally no more than 10~50HZ.
2.IMU
In fact, IMU is the core of the high-precision positioning and orientation system commonly known as INERTIAL navigation, which is also reflected in the price: for example, for a device worth nearly one million yuan, the satellite receiver only accounts for tens of thousands of yuan, and most of the rest are IMU costs.
To some extent, the selection of position and orientation system is actually focused on IMU, because the selection of GNSS is relatively simple and intuitive.
The rest of this paper also focuses on IMU.
IMU is usually composed of three single-axis accelerometers and three single-axis gyroscopes. The accelerometer detects the acceleration signal of the carrier in the coordinate system, while the gyroscope detects the angular velocity signal of the carrier relative to the coordinate system. After processing these signals, the attitude of the carrier can be solved.
It is important to note that the IMU and the inertial calculation algorithm provide relative raw location information. Its purpose is to measure the path relative to the starting point, so it does not provide specific information about where you are. Therefore, it is often used together with GNSS, when in some places where GNSS signal is weak or even missing, IMU can play its role, so that the carrier can continuously obtain the absolute position attitude information.
The update frequency of IMU is high, usually up to several hundred to 1KHz. Using three acceleration values, the displacement can be obtained by integrating twice to achieve position positioning, and the attitude information can be obtained by integrating angular velocity values, which can be combined to obtain the actual state of the object.
Although IMU is a strange technology, it can be used in mobile phones, cars, planes and even spaceships. The difference lies in materials, cost and precision.
According to different application scenarios, the precision of IMU has different requirements. High precision also means high cost. We use the highest class, of course.
IMU price and precision comparison
3. The odometer
The standard odometer is generally mounted on the wheel, and a rotary encoder is built in, which drives the common rotation through the wheel. The function is to measure the linear distance the vehicle moves and to help suppress drift errors in the event of a satellite losing lock. Odometer has a variety of forms, except hanging rotary encoder, magnetic grid, Hall and so on have applications.
4. Enhancement and auxiliary means
We mainly use some ground-based and satellite-based enhancement technologies, including RTK, RTD, PPK, PPP, DGPS, various SBAS, etc. Star-based enhancement systems are seldom used in our daily operations, and the ground-based enhancement systems are mainly used. High precision acquisition usually adopts the method of post-hoc differential decomposition.
Five, how is inertial navigation refined
As mentioned above, the positioning and orientation system includes inertial navigation. The main hardware of inertial navigation is IMU, which is composed of gyroscope and accelerometer.
1. The gyroscope
What does the top look like? Yeah, it’s this thing down here.
What does this gyro have to do with the angular velocity sensors that we understand in inertial navigation devices?
From the production process and material, principle, there is no big relationship, are two different things. But these things, can be used as a rotating angular velocity sensor, and this kind of similar mechanical gyro shape products used first, so later put this angular velocity sensor, are called gyroscope.
In fact, there are many kinds of gyroscopes, which can be generally divided into mechanical gyroscopes, laser gyroscopes and fiber optic gyroscopes, MEMS gyroscopes and so on.
Mechanical gyroscopes: used extensively in early aircraft. It has large volume, complex structure and poor precision. Later there are some improved types, such as: ball bearing free gyroscope, liquid floating gyroscope and so on.
Laser gyro: Measurement of rotational angular velocity using optical path difference in Sagnack’s theory. The point is that when a light beam travels in a circular channel, if the circular channel itself has a rotational speed, it takes more time to travel in the direction of the channel than it does in the opposite direction.
The Sagnac experiment
This is an interesting theory of Sagnac, and you can learn more about it yourself if you’re interested.
Laser gyro is actually a ring laser. In the closed optical path, the rotational angular velocity of the closed optical path can be measured by detecting the phase difference or the change of interference fringes when two beams of light from the same light source are transmitted in clockwise and counter-clockwise directions. When the ring laser is at rest, the optical path of the two lasers is equal, so the frequency is the same, the difference between the two frequencies (frequency difference) is zero, and the interference fringe is zero.
When the ring laser rotates about the axis perpendicular to the plane of the closed optical path, the optical path of the beam in the same direction is prolonged, the wavelength increases, and the frequency decreases. The other beam does the opposite, resulting in frequency differences and interference fringes.
Laser gyro has no internal motion device, low data drift rate, high reliability and high measurement accuracy. Due to the optical path difference, the laser gyro has a certain speed threshold below which angular velocity changes may not be detected. In addition, due to the use of ring laser, the whole device is large in size and high in cost.
Fiber optic gyroscope: strictly speaking, it also belongs to laser gyroscope, and the principle of laser gyroscope is the same, but with fiber instead of ring laser. The cost of optical fiber is low, but it is vulnerable to the uneven expansion and contraction caused by temperature changes and the change of tension during winding, so the accuracy is slightly lower. The laser gyro’s light propagates in the resonator, which is less affected by the outside world, so the precision is high, but the resonator is expensive. Because of its advantages in cost and volume, fiber optic gyroscope has been widely used in practice.
Micro mechanical gyro: MEMS gyro is a combination of micro mechanical structure etched by semiconductor technology and CMOS circuit technology. It is characterized by small size, low cost, easy to mass production, widely used in mobile phones, portable devices and other areas that do not require high performance. In addition, as technology advances, some high-end MEMS gyroscopes have become as accurate as fiber optic gyroscopes.
MEMS gyros work by sensing angular velocity using Coriolis forces, the tangential force exerted by a rotating object in radial motion. The vibrating object is suspended from the base by a soft elastic structure. The overall dynamic system is a two-dimensional elastic damping system in which vibration and rotation-induced Coriolis forces convert energy proportional to angular velocity into sensing mode and compute the output.
2. Accelerometer
The function of the accelerometer is to measure the acceleration force of the carrier to determine the position of the carrier in space and monitor the movement. Acceleration is a vector and the rate of change of velocity.
There are many types of accelerometers, including quartz flexible accelerometer, liquid floating accelerometer and MEMS.
According to the classification of sensing elements, generally can be divided into several types: piezoelectric accelerometer, piezoresistive accelerometer and capacitive accelerometer, in addition to thermal flow, resonant type.
Piezoelectric accelerometers use piezoelectric effects (piezoelectric materials generate electricity when subjected to physical stress) to sense changes in acceleration. Piezoelectric accelerometers are most commonly used for vibration and impact measurements. Piezoresistive accelerometers are much less sensitive than piezoelectric accelerometers and more suitable for vehicle crash tests. The resistance of a piezoresistive accelerometer is proportional to the pressure applied to it.
The most commonly used accelerometer is the capacitive accelerometer, which uses the change of capacitance to determine the acceleration of the object. When the sensor experiences acceleration, the distance between its capacitance plates changes with the movement of the sensor diaphragm, thus detecting the value of acceleration.
3. Why do integrated navigation
Practical applications, the single navigation model is hard to meet the navigation performance requirements, improve the overall performance of navigation system is an effective way to adopt integrated navigation technology, which USES two or more than similar navigation system for measuring and calculating the same navigation information, to form a measurement value, from the measured value is calculated for each navigation system error and correction.
For example, gyroscopes measure angular velocity, accelerometers measure linear acceleration. The former is the principle of inertia, and the latter is the principle of force balance. The accelerometer is correct in the long time, but has error in the short time due to the presence of signal noise. Gyroscope is more accurate in a short time, but there will be error in a long time. Therefore, the two need to adjust each other to ensure the correct signal.
For example, in the whole system, IMU provides carrier posture and can calculate the relative position. Its advantages are high update frequency, continuous output, stable data and good short-term stability, but its disadvantages are error accumulation, precision divergence over time and poor long-term stability, because position information and posture information are obtained by integration. The advantages of GNSS satellite navigation are that there is no position error accumulation and long-term stability is good, while the disadvantages are low update frequency and the possibility of discontinuous signal.
Therefore, GNSS and IMU can be complementary. Long-term absolute positioning can be achieved through GNSS, and IMU can be used to calculate positioning in the gap between GNSS position updates, and GNSS can be used to correct errors.
Since IMU and GNSS are complementary in performance, the combination of these two devices as a positioning system design is recognized as the best solution in the industry.
6. Main parameters of IMU
For performance parameter requirements, accelerometer and gyroscope need to be started from two aspects, there are many related indicators, generally mainly focus on the following:
7. Error factors affecting accuracy
In inertial navigation system, the hardware part mainly affects the accuracy of IMU, and its error sources are shown in the figure below:
Inertial navigation error source
1. Gyroscope influencing factors
As the core sensor of inertial navigation equipment, gyroscope plays an important role. Attitude calculation data largely depends on the data quality of angular velocity, so the accuracy of gyroscope will directly affect the quality of the solution. In other words, whether the IMU can correctly perceive the attitude of the carrier depends on the precision performance of the gyroscope.
2. Accelerometer influencing factors
In IMU, the influence of accelerometer is mainly reflected in the stability and accuracy of the accelerometer. The high precision of the accelerometer is to ensure the accuracy of subsequent data processing, and the stability of the accelerometer is one of the key factors directly affecting the normal performance of IMU.
3. Temperature influencing factors
In general, the working environment of inertial devices cannot be a constant temperature environment, especially the precision of gyro is seriously affected, so the influence of temperature can not be ignored.
4. Error processing
There are many error sources in the positioning and orientation system, and the errors in the hardware of inertial devices are generally divided into two types: systematic error and random error. The nature of systematic error is the error that can find the rule, so it can be compensated in real time, mainly including constant offset, scale factor, installation error, etc.
However, random error generally refers to noise, and it is difficult to deal with it because it cannot find appropriate relation function to describe noise. Allan variance, time series analysis and other methods are generally used for error modeling analysis of zero-offset data. For example, Kalman filter algorithm can be used to reduce the influence of random noise.
How to select a bit system
When designing a high precision acquisition system, an important thing is the selection of inertial navigation equipment, because it is not only related to the hardware cost, but also related to the precision performance of the final product. In specific selection, one of the main tasks is to investigate the IMU, no more than to pay attention to the following aspects, and then make a choice according to the product needs.
- Check whether sensor specifications meet requirements.
- Whether the price of the sensor is reasonable, whether the supply chain is complete.
- Whether the design difficulty of supporting hardware and software is acceptable.
- Whether the manufacturer’s technical support ability and supporting services are good.
Among them, the most important is the first and prerequisite, only when the technical performance meets the requirements, will go to the next cost, business considerations.
1. Index analysis
We still analyze from the most important technical aspects and introduce several key indicators in the process of gyro selection:
range
Range is in the selection of sensor must first determine when, the selected sensors used in what areas, generally for car, gyro choice within 300 degrees per second, accelerometers within 4 g is ok, other choice according to their own usage scenarios, such as airborne range to a few bigger, rail transit can be smaller. In the case of high precision, the precision will be higher if the range is smaller.
Zero bias and zero bias stability
In principle, drift occurs when the gyroscope is powered on or starts to work, which is divided into constant drift and random drift. Constant drift is called zero deviation, or zero drift, and the unit is °/h or °/s. By getting the gyroscope zero bias, we can in the subsequent use of compensation, but the compensation is the measured average for many times, in the output of the gyro constant drift after the compensation will be part of the residual, therefore, there is the zero of the gyro testing index of partial repeatability, it represent gyroscope zero degree of partial close to repeat every time, After calibration compensation, the residual constant drift of the gyro with good zero bias repeatability is small, and higher precision can be achieved.
Zero bias stability is obtained by calculating the variance of gyro output data in the process of primary electrification, and the constant drift mentioned above is deducted when calculating variance. Therefore, zero bias stability reflects the random drift index of gyro, also known as random noise.
Zero bias and zero bias stability, to a large extent, reflect the performance of gyro and accelerometer, it has long been regarded as a key indicator of inertial device specifications. Selection should be based on the cost and precision requirements, choose the appropriate model.
Angle random walk
When the gyroscope is in zero input state, the output signal is the superposition of white noise and slow variable random function. Diffuse random function can be used to determine zero bias and zero bias stability index.
The attitude and velocity length errors generated by gyro random walk are zero, but the velocity and attitude have certain oscillation errors, and the oscillation amplitude is related to the gyro drift random walk size. Random walk will cause position error with large amplitude, but the mean value of position error does not increase linearly with time, but presents a random walk process.
Random walk reflects the development level of gyroscope and the minimum detectable angular velocity of gyroscope.
Physical conditions
(1) Overall size
The evaluation of external dimensions mainly needs to select the appropriate size according to the actual vehicle installation situation, and the IMU should have good adaptability to the environment. For IMU body, its centroid position should be as close as possible to IMU physical center.
(2) Electrical and interface requirements
Conventional vehicle power supply is 9-16V, the selection should consider whether its working voltage range is consistent, otherwise it needs to add a power conversion module. In addition, it is necessary to examine whether there is a short circuit, overvoltage and other self-protection functions.
In terms of interfaces, you need to determine the interface type and cable type, such as USB, Ethernet, or serial port, and whether the data rate matches the communication mode to avoid data interruption and loss.
(3) Environmental requirements
Inertial sensors are very sensitive to temperature changes, so it is necessary to pay attention to its temperature drift related indicators, for some large temperature drift equipment, should pay attention to the installation environment, and try to provide a stable working environment temperature, such as adding fan/air conditioning heat dissipation, or heating device.
In addition, it is necessary to pay attention to the protection level of IMU, choose outside or inside the vehicle to install, and do the corresponding waterproof and dustproof measures.
(4) Relevant standards and specification requirements
Including but not limited to:
- General specification for vehicle-mounted satellite navigation equipment (GB/T 19392-2013).
- Automotive electronic equipment reliability test standard (ISO 16750).
- International Standard for Functional Safety of Road Vehicles (ISO 26262 2018).
- Environmental conditions and tests for electrical and electronic equipment of road vehicles (GB/T 28046).
2. Focus on testing
For inertial navigation equipment this kind of focus on performance, single function of the product, need to focus on the actual performance, can not fully believe the indicators in the technical manual, such as the following two commonly used factory mainstream equipment, you can see that the parameters are not very different, the accuracy is also very high, how is the actual performance?
Trimble POS LV 510 Nominal accuracy:
Novatel SPAN-CPT nominal accuracy index:
In fact, you get what you pay for, the price also naturally represents the performance, the accuracy of POS LV510 is actually much better than CPT, so the application scenarios of the two are different, 510 can do high precision acquisition, CPT is only used to do update or ADAS acquisition.
Also don’t be fooled by the specification, the nominal accuracy of 0.02 meters is the result of ideal operating conditions.
How else can you choose, other than indicators?
As the saying goes, a mule is a horse, and it is also suitable for selection. Our method is: actual measurement. After all, practice is the sole criterion for testing truth.
At present, positioning and orientation system level test methods mainly include: actual motion test, software simulation test, software – hardware combination of semi-physical simulation test. Among them, the actual motion test is obviously the most intuitive and real, and we usually use the sports car test to investigate the accuracy.
The test stage is generally divided into equipment preparation, route planning, data acquisition and data analysis.
Equipment preparation: Do the positioning precision of the directional system evaluation, need a reference standard, so that when driving the same route, the device under test trajectory and attitude error can get accurate quantification, practice is often with a high level of equipment as a benchmark, the true value, together with the equipment under test is installed in the same vehicle platforms, the track of the collection and the same route and carrier.
During installation, the part containing IMU needs to be firmly installed on the platform, and the measurement direction of the body should be orthogonal/parallel to the forward and horizontal direction of the vehicle, without free shaking; The installation position of the antenna should ensure that there is no block, the star receiving state is good, and there is no interference source nearby. Use odometer if necessary. Power cables and data cables are connected according to regulations.
Route planning: Different scenarios have different requirements for positioning systems, but for us, it is mainly road data acquisition, and it is the whole road condition acquisition, which is not only highway, country road, such as good satellite signal, not high requirements for inertial navigation. It also includes the scenes where the satellite signals of urban common roads, ring roads, viaduct up and down, parking lots and common national roads are covered or even interrupted. Therefore, the selection of test sections should be as comprehensive as possible to cover these scenes.
In addition to the requirements of normal test scenes, because the INS often have initialization requirements, so at the starting point and end point of the test line, open areas with good satellite signals should be selected.
Data collection: Data collection is relatively simple, and the data of the tested equipment and the true equipment can be collected according to the specified lines in accordance with the relevant operation specifications. In order to ensure reliability and consistency and facilitate comparison, the same line is often repeated acquisition. Reference station, you can set up their own, you can also use qianxun and other service providers of data.
Data analysis: After the completion of the sports car, the acquired inertial navigation and GNSS original data and reference station data are entered into the post-processing software for various pre-processing, format conversion, filtering, differential decomposition calculation, fusion, smoothing and a series of processes, and then the trajectory and attitude data are output.
The data analysis
Multiple tracks can be selected to compare with the output of true equipment in plane position, elevation, heading Angle, roll pitch Angle and other indicators, so as to obtain the error range in various scenarios. Based on this, the performance of the tested equipment can be analyzed to serve as the basis for selection.
In addition to using higher grade equipment as truth values to verify the accuracy of the equipment under test, other methods, such as calibrated close-range photogrammetry acquisition system + control field method, can also be used to estimate the accuracy.
Nine, summary
This paper introduces the basic situation of inertial navigation equipment and its sensors in positioning and orientation system. In fact, these sensors still have a lot of work to do in the application before they can be used by the end user. Such as the factory before the environmental test, aging, screening, turntable calibration and so on, is also very important, and the combination of calculation algorithm is the key, these combination of a set of boxing are laid, just have high precision.
Students who are interested in positioning and inertial navigation are recommended to read the series of books written by Teachers Qin Yongyuan and Yan Gongmin of NPU.