Real world design
Sensor fusion is a highly designed field that requires skillful masterming and simulation technology. It requires as much as possible to understand the working details of the sensor and their disadvantages and interactions. Over the years, people's concerns have been taken into navigation, smartphone applications and games. But until now, with a lot of knowledge, reserve and accumulation, people can get real and results.
In systems based on sensor fusion, the operation needs to be fine adjusted. There is nothing in the real world like "plug and play" is so simple. A system's trial operation requires adjustment parameters, and there is an interaction between the operations of each sensor, so it is easy to become a highly complex repetition process. Today's software has the ability to perform this "fine adjustment" at a high level, and you can provide a simple and intuitive filter adjustment program to OEM vendors (Figure 2).
Predefined filters make fine adjustment speed faster
Since the fine adjustment of the sensor fusion operation has developed and simplifies into a filter adjustment task, it provides developers with a valuable opportunity. By properly adjusting filters, developers or OEM vendors can allow the final product to operate in a market differentiation. Since all trade-off management is automated, developers can make effective decisions, such as overalllancing between stability and performance in order to adapt to the final target market.
Key Performance Index Measurement Settings
All sensor fusion technology is not equivalent. Different suppliers have a lot difference between existing technology and testing. In order to get the correct result, the correct software method containing the verified library must be used.
All hardware must be compatible and matched in terms of interfaces and timing parameters. The reasonable way is to ensure the performance of the camera system, which will move the orientation vector by tracking the mark on the object (this is a smartphone). The orientation vector is created with the sensor, and the vector recorded simultaneously with the data record is compared. Using this camera-based system allows for direct comparison of final business devices.
Static
Static is defined as a deviation between the device orientation and the actual device position when the device is placed in a stable position. In order to calculate static, it is necessary to collect the complete set of data in the heading, pitch and rollover when the phone is placed in multiple locations. The static main majority of the device is affected by the hardware parameters of the magnetometer and the gyroscope and the weights assigned to them. In a device with a low static value, the terminal user can see a large deviation in the heading of the compass or map application. When the device is in a static state, they can also see jitter in an interactive application (small rotation movement) . This is due to software correction gyroscope drift.
Dynamic
Dynamic is defined as the deviation between the device orientation between the device and the actual device position when the device is moved. Since the rotation acceleration is involved during exercise, it is more difficult to measure. Dynamic is calculated when the mobile phone moves in different motion modes (8-word dance, slow linearity, fast and slow rotation, and game action), by collecting tailoring, pitch and rolling. All data is collected in a fast possible data rate.
In a low-dynamic device, the terminal user can see that there is a lot of deviation between the movement and the actual motion of the screen on the screen. This is particularly attractive in enhanced reality applications because the movement of the enhancement unit is not synchronized. This is also one of the reasons why users feel unsatisfactory after using virtual reality a few minutes.
Although the direct relationship is not very obvious, the dynamics of large errors are also the main reason for the performance of indoor navigation applications. Since the user is navigating between known fixed points (such as starting from Wi-Fi or Bluetooth beacons), sensor data can be used to calculate trajectories. However, the heading error will accumulate over time, and therefore the device having a 15 ° poor dynamic device is easy to generate an accumulation error of more than 100 ° during the 20s ~ 30s time. Higher layer processing such as map match may be amended, but the cost is greater power consumption (Figure 3).
Calibration time
The calibration time is defined as a magnetic sensor in the calibration device in a pure magnetic field environment, which has never calibrated the time required to fully calibrate. All magnetic sensors need to be calibrated, but the method used for calibration defines whether the terminal user needs to calibrate and how to calibrate.
Some devices use a 8-word dance calibration method, that is, the terminal user will perform the device's calibration in the air. Even if you have experienced testers, this method also spends 5s ~ 6s time to complete the device calibration.
Device with shorter calibration time uses a gyroscope to calibrate the magnetic sensor, which means that calibration can operate in the background, and the required equipment movement is much smaller. These movements are usually performed in normal operation, and the terminal users never have to actively go to calibrate the sensor. Bosch Sensors Technology Company's Rapid Magnetic Calibration (FMC) algorithm is to use the latter method to ensure a shorter calibration time.
Directional stability time
Aggression stability time is defined as "post-motion" to reach the time required to stabilize a position. The orientation stability time should be as short as possible so that the user does not see the delay between the mobile device and the device stops and stabilizes between the correct position. This delay on the device is obvious when the static and dynamic of the device is very poor, because it takes more time correction errors in the movement. This effect is especially annoying in games and virtual / enhancement practical applications that need real-time response.
From a detailed assessment and analysis, it is apparent that the sensor fusion described herein can now be widely used in the level and consumer markets. Field trials show that users can have valuable upgrades in performance and aspects. Although hardware and software concepts and engineering technologies are more complicated, the tasks of transition from current sensors to this advanced solution from the current sensor are relatively simple.
Sensor fusion technology has now developed to a fairly mature stage. By designing the sensor and sensor fusion building module to the same package, it is possible to ensure that these units are obtained and can work well. System designers no longer need to spend time in assembly, optimization, and debugging traditional "forever online" subsystems, because through design, each device is optimized.
This highly technical and design creative brought the advantages of developers, you can provide OEM vendors to the huge benefits, but they can not only introduce highly differentiated products to the market, but also the entire new generation of electronic devices provided to users will also have significant Improved performance and efficacy.
Be
Source: Wiku Electronic Market Network
Our other product:
