Gyroscope
A gyroscope is a sensor that measures angular velocity, the rate at which an object rotates about an axis. In virtual reality and augmented reality hardware, the gyroscope is one of the core components inside an IMU (inertial measurement unit) and supplies the data used for rotational tracking of a headset or controller.
A classical mechanical gyroscope is a freely spinning disk mounted in a set of gimbals. Because a spinning mass tends to keep its axis of rotation fixed (conservation of angular momentum), the inner disk holds its orientation even as the outer frame is tilted or turned. This property let early gyroscopes act as orientation references in aircraft and ships. The devices used in consumer VR and AR are not spinning disks at all; they are tiny micro-electromechanical system (MEMS) gyroscopes etched into silicon, small enough to share a single chip with other sensors.[1]
What it measures
A gyroscope reports angular velocity, also called the rate of rotation. The output is an angular speed, usually given in degrees per second or radians per second, for rotation about each of the sensor's sensitive axes.[1] A three-axis gyroscope measures rotation about all three axes at once. In head and hand tracking these correspond to the three rotational degrees of freedom:
| Axis | Rotation | Head motion |
|---|---|---|
| Pitch | rotation about the side-to-side (lateral) axis | nodding "yes" |
| Yaw | rotation about the vertical axis | shaking "no" |
| Roll | rotation about the front-to-back axis | tilting the head toward a shoulder |
Because the gyroscope measures a rate of change rather than a fixed position, software has to integrate (accumulate) the angular velocity over time to estimate how far the device has turned. This integration is what turns a stream of rate readings into an orientation estimate for rotational tracking.[2]
MEMS gyroscopes
The gyroscopes in headsets, controllers, phones, and standalone VR devices are vibrating-structure MEMS sensors that work through the Coriolis effect. A small proof mass is driven to oscillate back and forth at high frequency along one axis, called the drive mode. When the chip is rotated, the moving mass experiences a Coriolis force that acts perpendicular to both its motion and the axis of rotation. That force pushes the mass sideways into a second oscillation, the sense mode, and the resulting tiny displacement is read out as a change in capacitance. The size of that change is directly proportional to the angular velocity, so the sensor turns rotation into an electrical signal.[1][3]
MEMS gyroscopes are well suited to VR and AR because they are small, light, inexpensive, draw little power, and start up quickly. The same advantages let manufacturers combine a three-axis gyroscope, a three-axis accelerometer, and sometimes a three-axis magnetometer into one compact package, the building block of the headset IMU.[1]
Role in the IMU and rotational tracking
Inside an IMU the gyroscope is the primary source of information about how the device is turning, and it carries most of the work for orientation tracking. The accelerometer measures linear acceleration and the magnetometer measures the local magnetic field, but only the gyroscope reports rotation rate directly. By integrating the gyroscope's three-axis output, the tracking system maintains a running estimate of the headset's orientation, which is then used to render the correct view as the user looks around.[2]
This rotation-only tracking gives the three rotational degrees of freedom (pitch, yaw, and roll) and is the basis of 3DOF tracking. Headsets that also track position in space (6DOF) combine the IMU data with additional sources such as cameras or external sensors, but the gyroscope still provides the fast, low-latency rotation signal at the heart of that fused result.[4]
Drift and sensor fusion
A gyroscope on its own cannot hold an accurate orientation for long. Each reading carries a small bias and noise, and because the readings are integrated over time, those tiny errors accumulate. The estimated orientation slowly wanders away from the true orientation, an effect known as gyroscope drift. It is most noticeable in yaw, the heading around the vertical axis, because nothing in the gyroscope alone provides an absolute reference to correct it.[2]
The standard fix is sensor fusion, blending the gyroscope with the other IMU sensors so each one covers the others' weaknesses. The accelerometer senses the constant pull of gravity, which gives a stable "down" direction and pins the pitch and roll of the estimate to the real vertical. The magnetometer senses the Earth's magnetic field, providing an absolute heading reference that pulls the yaw estimate back toward true and counters the drift the accelerometer cannot reach. Fusion algorithms such as complementary filters and Kalman filters merge these inputs continuously: the gyroscope supplies fast, smooth short-term rotation while the accelerometer and magnetometer supply the slow, drift-free long-term correction.[2][4] A device that carries all three three-axis sensors is often described as having a 9-axis IMU.[1]
Update rate
VR and AR gyroscopes are read at high rates to keep tracking smooth and latency low. Sampling the gyroscope more often produces a denser stream of angular-velocity values to integrate, which lowers orientation error and keeps the rendered view tightly synchronized with real head motion. The sensor in the Oculus Rift, for example, supported sampling rates up to 1000 Hz, reducing the delay between a head movement and the data reaching the game engine to roughly 2 milliseconds.[5] High update rates of this kind are a common reason dedicated VR headsets use purpose-built IMUs rather than relying on the lower-rate gyroscopes found in ordinary smartphones.
References
- ↑ 2.0 2.1 2.2 2.3 LaValle, Steven M.; Yershova, Anna; Katsev, Max; Antonov, Max. "Head Tracking for the Oculus Rift". https://msl.cs.illinois.edu/~lavalle/papers/LavYerKatAnt14.pdf.
- ↑ "Operation of MEMS Coriolis Vibratory Gyroscopes". https://www.azom.com/article.aspx?ArticleID=13265.
- ↑ 4.0 4.1 "Virtual reality headset". https://en.wikipedia.org/wiki/Virtual_reality_headset.
- ↑ "Building a Sensor for Low Latency VR". https://www.meta.com/blog/building-a-sensor-for-low-latency-vr/.