"With the powerful smart phones and other smart devices continue to progress, new applications emerging, new requirements system development often can not keep changing. Today, such as indoor navigation and augmented reality applications such as the use of new motion or position data are requires the user to accept was originally a simple game application development, sensor fusion is not perfect, but end users can easily find these implementations have considerable disadvantages and low.
http:: //www.eepw.com.cn/article/272212.htm address reference herein
Sensor Fusion is an innovative engineering, systems by integrating data from various sensors to ensure more complete and reliable sensor signal or the perceptual information. To achieve consistently sensor fusion, engineer before deciding how to optimize the integration of data from these sensors, in-depth understanding of the advantages and disadvantages of the sensor is very important. A successful method is to use a database based fusion accelerometer, magnetometer and gyroscope sensor signal and by compensating the disadvantages of each sensor, to provide high, stable and reliable orientation data.
With the end-users are being exposed to these new applications, and they hope to have a more reliable solution. Early indoor navigation sensor for GPS device between a known fixed track the user's position is very similar, only high-quality sensor fusion, can provide the desired fidelity, and the confidence of the user. OEM manufacturers are aware of this, most manufacturers believe that product differentiation is a good opportunity.
Another example is from virtual reality to enhance the reality of progress. In virtual reality (VR) system, the user is isolated from the real world, immersed in an artificial world. In augmented reality (AR) system, the user can still come into contact with the real world, while interacting with virtual objects around them. Use of existing technology, user information transfer delay is intolerable - this enhanced dislocation reality may lead to very poor user experience.
For OEM manufacturers and platform developers (ie, operating system developers), the challenge is the need to ensure that all these devices can provide stability to meet the performance requirements of the application. For example, Andrews devices have many different hardware and software combinations, each combination will produce different output quality. There is no standard test procedure, which means that application developers can not rely on Andrews sensor data to achieve the same performance on many different platforms. The following is a motion tracking camera system to analyze and compare the performance of different combinations of hardware and software so as to set the performance standards recommendations. Performance analysis is achieved by four key performance indicators measuring system (KPIs), these indicators are: static, dynamic, orientation stabilization time and calibration time.
Camera system generates a vector in accordance with the object orientation (smart phone) by the movement of the tracking marks on the object, and then compared with the vectors in the vector sensor phone created. Data logging applications simultaneously recording these vectors, to directly compare the application supports end-user device.
This paper introduces the concept of the smartphone environment sensor fusion, sensor fusion discuss how to use the software to improve overall, and describes a test method covers the measurement of performance results, including on the implementation of a number of flagship smartphones.
Fusion library article described the use of an accelerometer, magnetometer and gyroscope sensor signal compensation each other disadvantages and to provide a high, reliable and stable orientation data. The following focuses on the advantages and disadvantages of these key components, and how they are compensated with each other disadvantages (FIG. 1).
Figure 1: inherent characteristics of the MEMS sensor.
It describes the orientation of the object in three dimensions is how to place, in general, orientation with respect to a predetermined reference frame is defined in the coordinate system. Values of three independent at least as part of the three-dimensional vector used to describe the real orientation. All the points of the object will change their position during the rotation, the rotating shaft except the point.
Magnetometer
Magnetometer highly sensitive to local magnetic interference and distortion, which are likely to cause the calculated magnetic heading error. Such gyroscopes can be used to detect interference and heading change without rotating recording. By providing a sensor fusion magnetometer data is then compared to the gyroscope data more weight can be compensated.
The horizontal component of Earth's magnetic field used to calculate the magnetic heading, and pitch and roll angles, respectively along the horizontal (X) and vertical (Y) axis tilt angle. These affect the XY axis inclination angle of the magnetic field. When the device is not in a horizontal position, i.e., the inclination angle is not zero, calculate the heading will be incorrect. Therefore it is important to use an accelerometer, a position before the rotation of the plane XY is calculated by the inclination angle to ensure that these are compensated.
Accelerometer
Total acceleration value of the acceleration by the accelerometer can only measure movement of the device caused by the acceleration due to gravity, and including, the distinction between the two can not be detected, it is necessary to separate the gravity and motion:
Acceleration = linear acceleration - acceleration due to gravity
Gravity vector may be thought of as a three-dimensional vector indicating the direction and magnitude of gravity. When the device is in a stationary state, the output of a gravity sensor should be identical with the output of the accelerometer.
Linear acceleration can be seen as a three-dimensional vector indicating each device axial acceleration and gravity component is considered unrelated to the phone applications.
In this case it is obtained using the gyro. A gyroscope may be used to detect when the device is in a stationary state, and triggers the gravity vector is calculated offset calibration. The results can be used for dynamic sub-component (linear acceleration) during the movement of the computing device.
Gyro
A gyroscope may be provided around the three axes of rotation speed, it can be used to track the orientation of the device in motion. The rotational speed of the gyroscope can be tracked up to 2000 degrees per second (dps), and the speed can be tracked magnetometer is limited to about 400dps less. However, only the output of the gyroscope relative positions, it is necessary to have a distorted magnetic sensor as a reference.
All consumer gyros are inherent drift errors, even if the device is in a stable state, the gyro will change with time and a certain degree of temperature rotate. To correct these errors, can then be calculated with appropriate compensation accelerometer senses a stationary state high stability.
Real-world design
Sensor fusion is a design sophistication, the need to master the modeling and simulation techniques. It requires an understanding of the details of the work of sensors and their drawbacks and interaction situation as possible. Over the years, people's concerns have been brought into the field of navigation, smart phone applications and games. But until now, with the knowledge and accumulated large reserves, it makes it real and get results.
Based sensor fusion system, the fine adjustment operation requires. In the real world there is nothing like a "plug and play" so simple. A commissioning system parameters must be adjusted to the requirements, but there is no interaction between the operation of each sensor, it is easy to become highly complex iterative process. Today, this software has the ability to perform "fine adjustment" to high level, and may provide a simple and intuitive filter adjustment program (FIG. 2) to the OEM.
Figure 2: Typical sensor fusion software architecture.
Predefined filter that fine adjustment faster
Since the fine adjustment of sensor fusion operations have been developed and simplified to filter adjustment task, it provides developers with a valuable opportunity. By appropriate adjustment of filter, developers and OEM manufacturers can make the final products run to market differentiated way. Since all management is done automatically weigh, developers can make effective decisions, for example to make a trade-off between stability and performance, in order to adapt the final target market.
Setting key performance indicators to measure
Fusion is not the same for all sensors. In the existing implementation technology and testing, there is a big difference between different suppliers. In order to obtain correct results, the method comprising the software must be properly verified library.
All hardware must be compatible and match in terms of the interface and timing parameters. Reasonable approach is to ensure that the performance of the camera system, the system orientation vector generated by tracking an object in accordance with indicia on the object (in this case a smart phone) to move. Then the sensor orientation vector created, while recording the application data record vector compared. With this camera-based system allows direct comparison of the final commercial equipment.
Static state
Static is defined as a device in a stable position, the measured deviation between the actual device orientation device orientation. To calculate static, mobile phones need to collect data sets heading, pitch and roll of the plurality of terms in the position to stand still. Mainly due to a static device hardware parameters magnetometer and gyroscope and the right software to assign them were heavy impact. In the device having a low static value, the end user can see a large deviation in the heading the compass or the map application, when the device is in a static state, they can see the jitter interactive application (small rotational movement) . This is because the software calibration gyroscope drift.
Dynamic
Is defined as the difference between the dynamic device is in motion, the device orientation to the measured actual device orientation. Because during the movement comes to rotational acceleration, more difficult to measure. Is dynamic (slow linear, fast and slow rotation and operation of the game character dance 8,) during exercise is calculated by taking heading, pitch, and roll, etc. in different data sets in motion pattern of the mobile phone. All data at a faster data rate possible acquisition.
In the device having a low dynamic, the end user can see a great deviation between the actual motion of the mobile device on the screen. This is particularly noticeable in augmented reality applications enhance the mobile unit because the real world is not synchronized. This is the user in the use of virtual reality in a few minutes one of the reasons to feel dissatisfied.
Although the direct relationship is not very obvious, but also poor dynamic large error indoor navigation application performance of the main reasons. Since the navigation between the fixed point (for example, starting from the Wi-Fi or Bluetooth beacon) at a known user, the sensor data may be used to calculate the trajectory. However, the heading error will accumulate over time, thus having a 15 ° poor dynamic devices generate more than 100 ° easily accumulated error in time 20s ~ 30s. Such as map matching and other higher-level processing may be able to do some correction, but at the expense of greater power consumption (Figure 3).
Figure 3: departing from the azimuth change over time.
Calibration time
Calibration time is defined as the magnetic field environment in pure magnetic sensor calibration apparatus, the calibration status never make the time required for complete calibration status. All magnetic sensors need to be calibrated, the calibration method for the end-user defines whether calibration is required and how to calibrate.
Some devices employ 8 character dance calibration method, i.e., the end user will be prompted to make the device 8 to complete the calibration word exercise equipment in the air. Even if the test is done by experienced personnel, this approach would take 5s ~ 6s to complete device calibration.
The calibration apparatus has a short time using a gyro calibration of a magnetic sensor, which means that calibration can be run in the background, the amplitude of movement required by the device is much smaller. These movements usually carried out in a normal operation, the active end users never need to calibrate the sensor. Bosch Sensor Technologies rapid magnetic calibration (FMC) algorithm is to use the latter method to ensure a shorter calibration time.
Azimuth stabilization time
"After the movement" azimuth stabilization time is defined as the arrival time required for steady state position. Azimuth stabilization time should be as short as possible, so that the user can not see them stop moving device and the device stops moving and stability to the delay between the correct position. When the static and dynamic devices are very poor, it is obvious that the delay on the device, since more time correction error accumulated movement. This effect in the game need real-time response and virtual / augmented reality applications, especially annoying.
From the detailed assessment and analysis, it is clear the paper sensor fusion and now widely used in consumer-level market. Field test shows that the user can gain valuable upgrade in performance and respect. Although the hardware and software engineering concepts and more complex, but for developers, but it is relatively simple integration of the transition to this advanced solution of tasks from the current sensor.
Sensor fusion technology has now developed to a very mature stage. By constructing the sensor and sensor fusion module is designed into the same package, to ensure that these units can be obtained of, and be able to work well together. The system designer is no longer needed in the assembly, optimization and subsystems traditional "always on" time spent debugging, because by design, and power consumption of each device is optimized.
This highly technical and design creativity to bring the advantages of developers, can provide great benefits to OEM manufacturers, they are not only able to launch highly differentiated products to the market, and the whole new generation of electronic devices that provide users will also have a significant improved performance and efficiency.
Be
Source: WISDOM network "
Our other product:
