When you use sensors, you need to process the raw data into meaningful context. So, how do sensor hubs and sensor fusion work together to improve your product performance? Read on to find out.
Monday, April 29, 2019
Data can’t exist in a bubble. It is rendered meaningless if it can’t be interpreted to provide a clear picture of what it’s actually measuring. As described in our webinar (timestamp 19:36), sensor fusion is the process of fusing the raw data from multiple sensors together via algorithms to create one, coherent picture.
A sensor hub utilizes sensor fusion to turn raw sensor data into meaningful context. Sensor hubs are an idea that evolved from mobile devices: that sensor fusion could be done on a secondary, lower-power microprocessor to save power on the main processor. This enables better power management without sacrificing performance by allowing you to free up processor bandwidth for other tasks.
A look at sensor fusion:
Sensor fusion allows you to determine the state of your device by fusing data from multiple sensors. This is important because it allows you to cross-reference multiple sources of information, which improves the certainty of your data. For example, you can combine the long-term stability of an accelerometer and magnetometer with the short-term accuracy of a gyroscope to create a fluid estimate for orientation between static and dynamic motion. Sensor fusion enables more accurate, precise context by bringing multiple data sources together.
- Example of sensor fusion in action: An IMU (inertial measurement unit) blends the raw data from the motion sensors within a device in order to deliver full motion context, such as heading in a robot, activity classification for a wearable (walking/running/standing), or head orientation in a VR headset.
A look at sensor hubs:
A sensor hub is a central processing unit (CPU) or microcontroller with the explicit purpose of processing data from different sensors. This CPU is generally a lower-power device so that it can work independently from the main processor. Because it performs a specialized task, the sensor hub can save the device power that your main, higher-power CPU would otherwise have to spend to process the sensor data. A sensor hub is a great option for when you want to process data in the background, which allows you to run “always-on” sensors that use the low power sensor data.
- Example of a sensor hub in action: You can have an accelerometer running “always on” in the background via a sensor hub to count user steps in a pedometer, without drawing processing power away from your main application. It could also wait for significant, deliberate motion (such as a hard shake or distinct tap pattern) to wake up the main CPU to do more complex tasks.
If you’re curious about what you need to consider when using fused sensor data, check out the terminology you need to know in our post, Sensor Technology: Deciphering Your Choices. If you’d like to understand how we account for the most common sensor anomalies using sensor fusion algorithms, read our post, Universal Technical Challenges in IMUs, or check out our on-demand webinar for an overview of Using IMUs and Sensor Fusion to Unlock Smarter Motion Sensing.
We have more than 15 years of experience in developing sensors and sensor fusion, so if you have any questions or want to dig deeper into this topic, we are happy to help. Contact us!
Read more about CEVA’s SensPro™, the industry’s first high performance sensor hub DSP for processing and fusing data from multiple sensors including camera, Radar, LiDar, Time-of-Flight, microphones and inertial measurement units.
You might also like
More from Sensor fusion
Handling magnetic interference in sensors involves understanding your environment and developing algorithms to detect interference. A while ago, I covered a glossary …