Toggle search
Toggle menu
5.5K
6.1K
230
53.5K
VR & AR Wiki
Toggle preferences menu
Toggle personal menu
Not logged in
Your IP address will be publicly visible if you make any edits.

Positional tracking

vrarwiki.com · VR & AR Wiki
More actions
(Redirected from Positional Tracking)
See also: Tracking

Positional tracking estimates the location of a device or object relative to a reference frame. In virtual reality (VR) and augmented reality (AR), it is usually combined with orientation tracking to estimate a pose: position and orientation together. A rigid body has six degrees of freedom (6DoF), comprising translation along three axes and rotation about three axes.[1][2]

A 3DoF orientation tracker measures turning, tilting and rolling. Adding positional tracking lets a headset respond to leaning, ducking or moving sideways as well. The renderer can update the viewpoint to match these movements, including the changing perspective and motion parallax produced by head translation. Tracking a controller, hand or physical tool also lets an application place its virtual representation in the scene and interpret movement as an interaction.[2][3]

Tracking produces estimates with measurement error, delay and possible interruptions. Its usefulness depends on the task and environment: stable gameplay, alignment with a real tool and measurement of human movement impose different requirements. A system with low noise can still have an offset from the physical reference frame, as demonstrated in an independent study of the original HTC Vive.[4]

Reviewed 12 October 2026. Checked this article's pose definitions, tracking mechanisms, platform examples and measurement limits against the cited original papers and documentation. The embedded comparison has its own source review. About review dates.

Coordinate frames and pose

A position needs an origin, axes and units. A reported pose may locate a camera, a controller's grip point or a marker assembly, rather than the user's eyes or the tip of a tool. For example, the ZED SDK reports the pose of the stereo camera's left eye, while OptiTrack's Motive software reports a rigid body's pose at its defined pivot. Applications must use the appropriate relationship between that tracked frame and the point they want to display or measure.[1][5][2]

A local tracking origin need not correspond to a geographic location. OpenXR describes a pose as a relationship between two spaces at a specified time, rather than assuming one underlying global coordinate system. Its reference spaces include a viewer-relative VIEW space and a world-locked LOCAL space. In OpenXR 1.1, LOCAL_FLOOR places its origin at an estimate of floor height; the optional STAGE space has an origin on the floor at the center of a defined rectangular area.[6]

Recentering changes the relationship between application coordinates and the physical world. In OpenXR, a runtime that recenters LOCAL space must send a reference-space-change event with the time at which the new origin takes effect. A tracker can also revise its map or its estimate of a reference frame without the device physically moving. Applications that combine independently tracked objects must therefore account for their coordinate frames and the times of their pose estimates.[6]

Visual-inertial odometry estimates motion from camera and inertial measurements. In its usual formulation, global translation and the heading about gravity are unobservable without additional references. A local origin and heading can be chosen consistently without establishing an absolute geographic position. The same estimator can recover metric scale from the combination of image and inertial information under suitable motion and observability conditions.[7]

Methods of positional tracking

Markers on a Sensics dSight head-mounted display, illustrated in the 2014 tracking overview.
A visible optical fiducial used by InterSense, illustrated in the same overview.

Tracking systems use acoustic, radio, inertial, magnetic or optical measurements, sometimes in combination. Mechanical linkages can also measure pose within a constrained workspace. The sensing method and the placement of sensors are separate design choices: optical tracking, for example, can observe markers from outside the tracked object or use cameras carried by it.[8] The illustrations show examples discussed by Boger in 2014.[3]

Acoustic tracking

Acoustic tracking can estimate distance from the travel time of an ultrasonic pulse between an emitter and receiver. With known signal timing and propagation speed, a measured travel time supplies a range constraint. Measurements involving several known reference locations can constrain position; several receivers or emitters fixed to a rigid body can provide constraints on its orientation too.[2][3]

Room echoes and blocked direct paths can complicate pulse detection. Propagation speed varies with air conditions, and reverberation can restrict how quickly separate acoustic measurements can be made. These limitations depend on the design and installation. In a hybrid tracker, inertial measurements can update the pose between acoustic observations.[8]

The InterSense IS-900 is an example of inertial-ultrasonic 6DoF tracking for simulators and immersive displays. Its manufacturer distinguishes wired and wireless specifications and states that published accuracy was measured in a manufacturing laboratory. The product page also records a May 2025 end-of-life announcement and a last-time-buy period through December 2025 or until supplies lasted; it describes a historical commercial system rather than an unrestricted current purchase option.[9]

Wireless tracking

Radio positioning commonly uses tracked tags and reference anchors. Ultra-wideband (UWB) systems can use two-way ranging, which exchanges messages to estimate propagation time, or time-difference-of-arrival measurements involving anchor pairs. These measurements constrain a tag's position relative to the anchor layout. Distance-based location estimation is commonly described as trilateration or multilateration; angle measurements support triangulation.[10][2]

UWB does not have one universal positioning accuracy or update rate. Anchor geometry and non-line-of-sight propagation affect the estimates. Zhao and colleagues tested anchor placement with DW1000 radios in a cluttered indoor space using Vicon for reference positions. Their results showed that the placement of anchors changed localization error, and their model explicitly included biases caused by obstacles.[10]

IndoTraq's September 2018 location-based VR white paper describes combining UWB/IMU tracking with inside-out camera tracking. The radio-inertial subsystem provides a reference position and assists camera tracking when it drifts or loses observations. The paper's precision and delay figures are manufacturer claims for its described systems; they are not specifications for UWB tracking in general.[11]

Inertial tracking

An inertial measurement unit (IMU) typically combines accelerometers and gyroscopes. A gyroscope measures angular velocity. An accelerometer measures specific force in the sensor's frame; obtaining translational acceleration requires accounting for orientation and gravity, along with sensor bias. Integrating acceleration updates velocity, and integrating velocity updates position from an initial state.[7]

Inertial sensors can supply frequent motion measurements without seeing external markers. Their estimates accumulate drift from bias, noise and errors in orientation or gravity compensation. A position estimate based solely on integrated acceleration does not supply a lasting absolute position reference. Cameras, acoustic ranges or other observations can constrain that accumulating error.[2][7]

A sensor's number of measured axes does not by itself establish the degrees of freedom of the pose a system can track reliably. For example, adding a magnetometer gives magnetic-field measurements; using Earth's field to correct heading does not establish the device's position. A 3DoF headset may use a neck model to approximate eye displacement during head rotation, but this model cannot measure independent translation of the head or torso.[2]

Magnetic tracking

Electromagnetic tracking uses a deliberately generated magnetic field and sensors that measure its local properties. With a known source arrangement, measurements from differently oriented source and receiver coils can constrain position and orientation relative to the source. Designs can use alternating-current or pulsed direct-current excitation. This differs from sensing Earth's magnetic field for heading correction.[8][3]

Magnetic tracking can work without an optical line of sight. Its working volume and errors depend on field strength, distance and distortion from the environment. Nearby conductive or ferromagnetic material can alter the field; field compensation is a design-specific response to this problem. Polhemus, for example, describes AC electromagnetic trackers and separate technology intended to compensate for metallic distortion.[12][13][8]

The Razer Hydra controllers are a consumer example. Razer's April 2011 announcement described six-degree-of-freedom magnetic motion tracking with a low-power field generated by the base station. Its published precision and latency statements were product claims, not comparative evidence that magnetic tracking is superior to every other technique.[14]

Optical tracking

Optical tracking estimates pose from received light. Camera systems detect image features, while some other optical systems use photodiodes to detect timed illumination. Visible markers, infrared markers and natural scene features provide different kinds of reference. Occlusion and the visibility of those references constrain tracking.[2][15]

Tracking with markers

A marker arrangement attached to a rigid object supplies a known geometry. Cameras observe the markers, and the tracking software estimates the object's pose from their image or reconstructed positions. The geometry must remain fixed. Distinct arrangements help the system distinguish several tracked objects; additional markers can leave useful observations available when some are hidden.[5]

Passive retroreflective markers return illumination toward its source. Active markers emit light, typically from LEDs. OptiTrack's Motive 3.0 documentation supports both passive and active marker rigid bodies and requires at least three markers to create one. Its recommendations about marker layout and identity apply to that software; other camera-pose methods have their own geometric constraints.[5][2]

Tracking with visible markers

Fiducial markers provide recognizable patterns with known dimensions. A camera can detect a marker's corners and identity, then estimate its pose using the camera's calibration and the marker's geometry. In OpenCV's ArUco workflow, the Perspective-n-Point calculation returns rotation and translation between the marker and camera coordinate systems. The marker's specified size sets the scale of the translation result.[16]

The visible InterSense fiducial in the illustration is another example. Markers can be installed in the environment for a moving camera to observe, or attached to an object seen by a stationary camera. Camera calibration, the visible marker geometry and missing observations affect the pose solution.[3][2]

Markerless tracking

Markerless tracking uses image information without requiring purpose-built fiducials. One approach matches an object's visible features to a known geometric model. Another tracks natural features of the surroundings and estimates the motion of the camera itself. These tasks use computer vision, but estimating a camera's pose in a room is a different task from recognizing and following a particular object.[3][17]

Simultaneous localization and mapping (SLAM) estimates the sensor's pose while building a map. In ARCore, visually distinct image points are tracked and combined with IMU measurements to estimate camera pose. In Oculus Insight, repeated observations of room landmarks help constrain headset drift, while headset cameras also detect infrared LEDs on the controllers. Thus a markerless headset and marker-based controllers can belong to the same tracking system.[17][18]

Depth map tracking

A depth camera supplies distances to visible surfaces. These measurements can support tracking in more than one way. The original Kinect body-pose research classified pixels of a depth image into body parts and generated confidence-weighted 3D joint proposals. This estimated the person relative to the camera; it did not automatically locate the camera in a room.[19]

KinectFusion addressed camera tracking and reconstruction instead. The 2011 system aligned incoming depth frames with a fused surface model using iterative closest point (ICP) estimation. Its authors described tracking failure when motion was too large for the pose update or the visible geometry did not constrain all six degrees of freedom. The implementation used interactive relocalization, asking the user to align the live depth image with a prediction from the last known pose.[20]

Sensor fusion

Sensor fusion combines observations with different strengths and failure conditions. A visual-inertial system can propagate pose using fast inertial measurements between camera frames, then constrain that estimate with visual observations. Implementations can use filtering or nonlinear optimization; the appropriate estimator depends on the model and the available measurements.[7]

The sensors require compatible coordinate frames and timing. Forster and colleagues' visual-inertial estimator explicitly models the fixed camera-to-IMU transform and the IMU's gyro and accelerometer biases. It preintegrates several inertial samples between keyframes into a motion constraint. These are components of an estimation method, not a guarantee that fusion removes error or permits indefinite tracking during visual occlusion.[7]

Oculus Rift and HTC Vive's positional tracking

The original Oculus Rift and HTC Vive illustrate two different optical arrangements. Rift's Constellation uses external camera observations of infrared LEDs on the tracked hardware. In the earlier Oculus Rift DK2, the global-shutter tracking camera was synchronized with headset IR illumination, and the headset also collected inertial data. Oculus documented the DK2 hardware design in its October 2017 open-source release.[2][21]

Vive's Lighthouse system uses fixed base stations that emit sweeping infrared light. Photodiodes on the headset and controllers detect that light, and timing measurements constrain their poses. The base stations provide an optical reference; they do not observe the user through cameras. Inertial measurements are combined with the optical observations.[15][2]

Coverage depends on which observations remain available. Oculus's 2019 account of Constellation-tracked controllers explains that a controller can leave a camera's view or have its tracking ring occluded. The original Vive study likewise observed tracking loss when the headset could not see either Lighthouse emitter. These observations do not establish the same fallback behavior for every later headset or software revision.[22][4]

Positional tracking and smartphones

Phones can estimate translation as well as rotation. ARKit combines motion sensing with camera-image analysis in its world-tracking configuration; ARCore likewise combines visual feature tracking and inertial measurements. Their local tracking frames let virtual objects respond to device translation. This capability depends on the supported device, the session configuration and the tracking conditions.[23][17]

Phone-based world tracking still has limits. Apple's documentation identifies dark scenes, insufficient visual detail and excessive motion as conditions that reduce tracking quality. An orientation-only session or a neck model has a different purpose and does not become independent 6DoF position tracking simply because it produces an approximate eye location.[23][2]

The 2018 ADVIO study compared ARKit on an iPhone 6s with iOS 11.0.3, an ARCore developer preview on a Google Pixel and Tango on a dedicated tablet. It used a rigid capture rig and inertial reference trajectories constrained by manually identified positions across indoor and outdoor routes. Those position fixes were modeled with 25 cm standard deviation, rather than treated as exact coordinates. The researchers observed drift and failures, including difficulty reconciling the moving device with a static camera view in a closed elevator. Those results describe the tested historical hardware and software, rather than a ranking of today's mobile tracking platforms.[24]

Types of positional tracking

The terms inside-out and outside-in describe sensing arrangements, while marker-based and markerless describe what the system observes. The categories can be combined. Calling a system inside-out also does not necessarily mean that it has no installed reference hardware.[8][2]

Arrangement Description Examples and distinctions
Inside-out tracking Sensors on the moving device observe references in the surroundings. Headset cameras can observe room features, or a camera can observe installed fiducials. Lighthouse has photodiodes on the tracked devices and fixed emitters; authors classify it differently, so the sensor arrangement is more informative than the label alone.[8][4]
Outside-in tracking External sensors observe the moving target. Rift Constellation cameras observe target-mounted IR LEDs. Calibrated motion-capture cameras can track marker assemblies on an HMD, controller or tool.[2][5]
Markerless inside-out tracking A moving camera tracks natural scene information without installed fiducials. Oculus Insight estimates headset pose with camera and IMU data. Its controller tracking still uses LEDs.[18]
Markerless outside-in tracking An external camera estimates an object's or person's position from visible appearance or depth information. The Kinect body-part method estimates 3D joint proposals from the person's depth image, without target-mounted optical markers.[19]

Sony's PlayStation VR2 is another documented headset-camera arrangement: its specifications list four embedded cameras for headset and controller tracking, with IR LEDs and six-axis motion sensors in the Sense controllers. Headset tracking and controller tracking therefore need separate descriptions even when both use the same cameras.[25]

Drift, relocalization and tracking loss

A pose estimated by accumulating motion can gradually diverge from its reference. Loop closure uses observations of a previously visited place to constrain that accumulated error. Relocalization attempts to identify a current pose within an existing map after an interruption or loss of tracking. As one implementation example, StereoLabs' ROS 2 wrapper describes an odometry frame based on visual-inertial odometry and a map frame that applies loop-closure corrections. It also provides an option to reset odometry when loop closure occurs.[1][26][27]

Spatial anchors let applications associate content with a tracked location. They do not freeze every numerical coordinate permanently. ARCore can revise trackable poses as its environmental estimate improves, and an anchor maintains the relationship of attached content to its tracked location. Apple's world-map relocalization can restore saved anchors, but success depends on recognizing the environment again; a session can remain in its relocalizing state if it cannot do so.[17][27]

A pose value can remain available while its position is no longer actively observed. In OpenXR, position validity and active positional tracking have separate flags. A runtime can return an inferred or last-known position with XR_SPACE_LOCATION_POSITION_VALID_BIT set and XR_SPACE_LOCATION_POSITION_TRACKED_BIT clear. Applications must inspect these flags and must not read a position whose validity bit is unset. The flags describe the relation between the requested spaces, including special cases for spaces tied to the same entity, rather than a universal health indicator for the entire tracking system.[6]

Fallback behavior is platform-specific. Google's historical WorldSense documentation describes reverting to 3DoF when insufficient features, darkness or complete camera occlusion prevent positional tracking, then recovering when conditions improve. Microsoft's Windows Mixed Reality controller documentation describes a short period of inertial position tracking outside the headset cameras' view, followed by a body-locked approximate position; in some circumstances it provides no position. These are documented behaviors of those platforms.[28][29]

Evaluating a tracking system

A tracking figure needs a defined quantity and measurement procedure. Spatial accuracy describes agreement with a reference; precision describes the variability of repeated measurements. Tracker delay also differs from the full motion-to-photon latency of the sensing, rendering and display pipeline. An update rate alone does not establish that delay.[4][8]

Quantity or condition What to establish
Position and orientation error Which reference system and coordinate registration were used; which points on the device were compared; the tested positions and motion.[4]
Jitter and drift The variability of a stationary pose versus accumulating or slowly changing error. Low jitter can coexist with a biased coordinate frame.[4][2]
Timing Sensor timestamp, pose output time, prediction target and display response. Historical and predicted poses refer to different requested times in OpenXR.[8][6]
Coverage and recovery Which sensors or markers are visible, where failures occur, and whether the system reports observed, inferred or unavailable position during recovery.[22][6]
Environment and configuration Lighting, visual texture, obstacles and marker or anchor placement; the exact device and software used.[23][10][24]

In the original Vive study published in 2017, Niehorster and colleagues measured low sample-to-sample noise but also a reference plane tilted from the physical floor and offsets after tracking was lost and regained. Their approximately 22 ms end-to-end latency estimate was an upper bound from a display-response experiment. Neither the jitter figure nor that delay is a general accuracy specification for every SteamVR system.[4]

Comparison of tracking systems

See also: Comparison of tracking systems

The tracking systems below include historical console motion controls and tracking used by VR systems. Their sensing arrangements and published performance figures differ.

The systems below differ in where the sensors and the reference points sit (outside-in or inside-out), in whether they follow purpose-built markers such as LEDs and photodiodes or natural features of the room, and in how they combine optical measurements with an inertial measurement unit (IMU).[30] Early consumer systems such as the Wii Remote and PlayStation Move were built for console games;[31][32] the Oculus Rift DK2 developer kit added camera-based positional tracking in 2014, and later headsets moved the cameras onto the headset itself.[33][34]

Published accuracy and latency figures for these systems mix manufacturer claims with laboratory measurements made under different conditions, so a number for one system is often not directly comparable with a number for another. Welch and Foxlin point out, for example, that a tracker's update rate contributes to its latency but "doesn't tell the entire story", because the delay also depends on the length of the processing pipeline.[30]

Reviewed 12 October 2026. Reopened the cited manufacturer, reporting and original academic sources for the table and tracking descriptions. Corrected product-scope errors, attributed vendor figures and bounded the experimental results to their tested configurations and methods. About review dates.

Comparison table

The table compares the main tracking system of each product. The optical-reference column identifies light or markers, rather than an optical carrier frequency. The spatial column mixes attributed accuracy claims and explicitly labeled jitter measurements; these quantities are not interchangeable. Where a manufacturer figure or an independent measurement is available it is cited in the cell; the section Published accuracy and latency measurements gives the conditions behind the measured values.

Brand & Model Tracking system Inside-out Outside-in Marker-based Optical
reference 		Inertial
sensing 		Published spatial
figure (mm) 		Latency
(ms)
Facebook/Oculus Rift Constellation No Yes[35] Yes Infrared[36] Yes[35] ? ?
IndoTraq HSVT Yes[37] No No None (camera tracking plus an ultra-wideband radio tag)[38] Yes[38] Sub-millimeter (2018 vendor claim; no measurement method stated)[38] 10 (2018 vendor comparison claim; no measurement method stated)[38]
HTC Vive (original)/SteamVR Tracking 1.0 Lighthouse Yes[39] No Yes Infrared[39] Yes[39] Below 0.2 (RMS jitter, measured)[39] About 22 (end-to-end, measured upper bound)[39]
Microsoft HoloLens Inside-out head tracking (4 environment understanding cameras, 1 depth camera)[40] Yes No No None Yes[40] ? ?
Nintendo Wii Remote Infrared camera in the remote and Sensor Bar[31] Yes No Yes Infrared[41] 3-axis accelerometer[31] ? ?
Sony PSVR PlayStation Camera[42] No Yes Yes Visible light: blue headset lights;[42] color-changing sphere on PlayStation Move[32] Yes[42] ? Less than 18 (Sony figure for the 2015 prototype)[43]
Google WorldSense Yes[44] No No None Yes[45] ? ?
Oculus Quest / Oculus Rift S Oculus Insight Yes[46] No Controllers only Infrared (controller LEDs)[46] Yes[46] Sub-millimeter (Facebook design target)[46] ?
Windows Mixed Reality Inside-out tracking by headset cameras[47] Yes No Controllers only LEDs on controllers[47] Yes[47] ? ?
PlayStation VR2 Inside-out tracking by 4 embedded cameras[48] Yes No Controllers only Infrared (controller LEDs)[48] Yes[48] ? ?

A question mark means that no applicable figure was established in the sources checked here.

Tracking approaches

Outside-in and inside-out

When optical emitters and sensors are used for tracking, the designer has to decide whether to put the light sources on the moving object and the sensors in the environment, or the other way round. Welch and Foxlin, in a 2002 survey of motion tracking in IEEE Computer Graphics and Applications, note that the first arrangement is usually called outside-looking-in and the second inside-looking-out, but warn that the labels "can be misleading". Their example is a measurement system with optical sensors on the target and spinning light sources in the room that sweep planes of light across them: the sensors look outward, yet the system has the orientation sensitivity of an outside-in design. In their view, the real distinguishing factor is whether the bearing angles to reference points are measured from the outside or the inside.[30]

Valve's Lighthouse system works in the same swept-light way, so it is classified differently by different writers. Niehorster, Li and Lappe describe the HTC Vive tracker as operating on "a so-called inside-out principle, where no external cameras are needed", because the photodiodes on the headset and controllers do the sensing while the two base stations only emit light.[39] The table above follows that description.

An outside-in camera system, such as Constellation on the Oculus Rift or the PlayStation Camera used with PlayStation VR, observes object-mounted tracking lights with cameras placed in the environment. Oculus recommended additional sensors for 360-degree and room-scale use, because the user's own body could block the LEDs from a single camera's view.[35] Inside-out camera systems such as Oculus Insight, WorldSense and the tracking in PlayStation VR2 track the headset against the room itself without external tracking cameras or beacons.[46][34]

Markers

Marker-based systems follow reference points that the manufacturer builds into the hardware. These can be active light sources, such as the infrared LEDs of Constellation and the Wii's Sensor Bar or the visible lights on the PlayStation VR headset, or sensors, such as the photodiodes that Lighthouse base stations sweep with lasers.[36][41][42] Markerless inside-out systems instead use natural features of the environment. Oculus Insight is built on visual-inertial simultaneous localization and mapping (SLAM), which fuses camera images with IMU data to fix the headset's position within a map of the room that is updated continuously.[46] Oculus Quest, Rift S and PlayStation VR2 combine the two approaches: the headset tracks itself without markers, while the hand controllers carry infrared LEDs that the headset's cameras follow.[46][48]

Welch and Foxlin identify the main weakness shared by all optical systems as the need for "a clear line of sight between the source and the sensor".[30]

Inertial sensing and sensor fusion

Most systems in the table also carry inertial sensors (gyroscopes and accelerometers) in the tracked device. Inertial sensors need no emitters in the room and have very low latency, but they drift: Welch and Foxlin calculate that an accelerometer bias of just 1 milli-g would make a position estimate drift by 4.5 meters after 30 seconds.[30] Other observations constrain the accumulating error through sensor fusion. The sensor that Oculus built for its 2013 Rift prototypes tracked head orientation only: it sampled at up to 1,000 Hz and used the accelerometer's estimate of the "down" vector and a magnetometer to correct the gyroscope's drift.[49] The DK2 added an array of infrared LEDs and a near-infrared CMOS camera running at 60 Hz for position, alongside a gyroscope, accelerometer and magnetometer updating at 1,000 Hz.[33]

In their 2017 description of the original HTC Vive, Niehorster and colleagues reported that IMU dead reckoning supplies frequent pose updates and Lighthouse observations constrain accumulated error at 120 Hz. In the tested system, loss of both base stations caused the reported position and orientation to stop updating.[39] Windows Mixed Reality controllers behave differently: when a controller leaves the headset cameras' field of view it keeps returning high-accuracy poses for a short time from inertial tracking, then falls back to an approximate position locked to the user's body while still reporting its true orientation.[47]

Other sensing methods

Optical tracking is not the only option. Ivan Sutherland's 1968 head-mounted display (Sword of Damocles) used a mechanical tracker made of a telescoping arm with a universal joint at each end, supplemented by a continuous-wave acoustic tracker that measured the phase shift of the received signal.[30] Acoustic trackers have their update rate limited by room reverberation, and continuous-wave designs suffer from multipath reflections. Magnetic trackers need no line of sight because magnetic fields pass through the human body, but nearby metal distorts them. Welch and Foxlin's analysis relates radial positional jitter to the fourth power of source-sensor distance.[30]

Electromagnetic tracking was used in the Control controller of the Magic Leap One AR headset. According to its 2018 FCC filing, reported by Road to VR, the controller contains "a transmitter that generates 3 orthogonal AC magnetic fields at frequencies ranging from 28.5 kHz to 42.42 kHz", which gives it six degrees of freedom.[50]

Radio positioning provides another tracking arrangement. IndoTraq introduced a tracking tag at CES 2016 that combined an IMU with ultra-wideband (UWB) radio positioning; the company quoted a precision of plus or minus 5 mm and an update rate of 200 Hz.[51]

Systems in the table

Oculus Rift (Constellation)

Constellation uses an array of infrared LEDs on each tracked device, followed by an external camera.[36] Oculus's 2017 DK2 hardware account calls that developer kit's LED modulation an early implementation of Constellation.[52] The March 2014 DK2 announcement reported submillimeter positional accuracy, a 60 Hz camera and 1,000 Hz inertial updates; those DK2 figures are not consumer Rift specifications.[33] The Oculus Touch controllers use outside-in LED tracking; an October 2016 report described their scheduled 6 December launch. Oculus's recommended layouts at the time included two sensors on opposite sides for standing 360-degree play and three sensors in a triangle for room-scale play.[35]

IndoTraq (HSVT)

IndoTraq's High Speed Vision Tracking (HSVT) combines inside-out camera tracking on a headset with the company's HSKT tracking tag, which fuses UWB radio ranging with IMU data. In the company's 2018 white paper for location-based VR, IndoTraq claims 5 mm precision and 5 ms measurement latency for the UWB/IMU tag, which supplies a reference position and drift correction for camera tracking. The paper illustrates a combination with HTC Vive Focus tracking. These subsystem figures differ from the fusion-system claims in its comparison table and have no stated measurement procedure.[38][37] IndoTraq describes HSVT as having sub-millimeter precision, but its own pages give different update rates: 300 Hz on the virtual reality use-case page and 100 Hz on its HSVT product page when checked on 12 October 2026. The pages do not establish a common test configuration that resolves the difference.[37][53]

HTC Vive and SteamVR (Lighthouse)

In the original HTC Vive, two base stations alternately send out horizontal and vertical infrared laser sweeps spanning 120 degrees in each direction. Photodiodes on the headset and controllers register when the laser hits them, and the differences in hit times give the device's position and orientation.[39] Valve's developer page offers SteamVR Tracking without licensing fees. Its description of the original sync-pulse arrangement lists base stations that sweep the room with sync pulses and laser lines to about 5 meters, up to 32 sensors per tracked object, a 1,000 Hz IMU, tracking at 250 Hz to 1 kHz, and a vendor claim of "sub-millimeter accuracy".[54] The second-generation base stations sold with the Valve Index use a single rotor; Valve lists a 160 by 115 degree field of view, a range of 7 m, and support for up to four base stations covering a play space of up to 10 m by 10 m.[55]

Microsoft HoloLens

The first-generation HoloLens carries its tracking sensors in the visor. They include one IMU, four environment understanding cameras and one depth camera. Microsoft's documentation for the device is now in its archived previous-versions library.[40]

Nintendo Wii Remote

The Wii Remote contains a 1024 by 768 infrared camera with built-in hardware blob tracking of up to four points at 100 Hz, and a 3-axis accelerometer that also operates at 100 Hz.[31] The camera follows the Sensor Bar, which has five infrared lights on the front of each side.[41] Johnny Chung Lee showed that the same camera could be used for head tracking in "desktop VR" by pointing a Wii Remote at a pair of infrared LEDs worn on the head.[31]

Sony PlayStation VR

PlayStation VR relies on the PlayStation Camera, which captures the position, angle and movement of the headset; the headset's tracking lights glow blue and it contains a 6-axis motion sensing system (3-axis gyroscope and 3-axis accelerometer).[42] At GDC 2015 Sony said that its revised Project Morpheus prototype had nine LEDs "to support robust 360 degree tracking" and a latency of less than 18 ms, about half that of the first prototype.[43] The PlayStation Move controllers, announced for PlayStation 3 in March 2010, combine a three-axis gyroscope, a three-axis accelerometer and a terrestrial magnetic field sensor with a color-changing sphere that is tracked by the camera.[32]

Google WorldSense

WorldSense is Google's inside-out positional tracking system. Google's archived documentation describes combining visually distinctive environmental features from dual fisheye cameras with measurements from the headset's IMU.[45] On the Lenovo Mirage Solo it uses a pair of cameras on the front of the headset for 6DoF head tracking, while the included controller is limited to 3DoF rotation. In its May 2018 review, Road to VR called the headset's positional tracking "robust, accurate, and low latency".[44]

Oculus Insight

Oculus Insight, introduced with the Oculus Quest and Oculus Rift S in 2019, uses ultra-wide-angle cameras in the headset, IMUs in the headset and controllers, and infrared LEDs in the controllers that the headset cameras detect. Facebook engineers wrote that the system computes a position for the headset and controllers every millisecond, that its precision goal was in the sub-millimeter range, and that it was validated against OptiTrack motion capture systems installed in Facebook workspaces and employees' homes, across hundreds of environments with different lighting, decorations and room sizes.[46][56]

Windows Mixed Reality

Windows Mixed Reality headsets and their motion controllers need no external tracking sensors; the headset tracks the controllers optically.[47] Microsoft announced in December 2023 that Windows Mixed Reality was deprecated, and Windows 11 version 24H2 removed support entirely. Microsoft said existing devices would keep working with Steam through November 2026 on Windows 11 version 23H2.[57]

PlayStation VR2

Sony announced PlayStation VR2 for global availability on 22 February 2023. Its tracking arrangement replaces the external camera of the first PlayStation VR. Four cameras embedded in the headset track both the headset and the PS VR2 Sense controllers, which are followed through a tracking ring across the bottom of each controller.[34] Sony's specifications list a six-axis motion sensing system in both the headset and the controllers and IR LEDs in the controllers for position tracking.[48][58]

Published accuracy and latency measurements

Welch and Foxlin separate tracker error into several quantities: spatial distortion (repeatable errors across the working volume), spatial jitter (noise while the tracker is still), creep (slow changes over time), latency, latency jitter and dynamic error during motion.[30] Independent studies usually measure only some of these, with their own reference systems and room setups.

Study Systems tested Setup Main findings
Niehorster, Li and Lappe (2017)[39] HTC Vive (first-generation Lighthouse) Two first-generation Vive systems; physical floor grid with placement uncertainty, Vizard 5.5 pose polling at 90 Hz; WorldViz PPT-X plus InertiaCube comparison and 480 Hz video/display latency test Sample-to-sample RMS jitter below 0.02 cm and 0.02 degrees; end-to-end latency of about 22 ms (an upper bound); reference plane tilted relative to the floor, with large offsets after tracking was briefly lost
Holzwarth, Gisler, Hirt and Kunz (2021)[59] Oculus Quest 2 (Oculus Insight); SteamVR Tracking 2.0 with HTC Vive Trackers Stationary measurements at 36 points on a 5 by 5 m grid, four Base Station 2.0 units, Quest 2 via Link and 2018 Vive Trackers on a carriage; 50 Hz pose sampling for 20 s per point Quest 2 had significantly lower height error and lower sample-to-sample position jitter in this static setup; the comparison used the better-performing of two Vive Trackers after outliers in the other
Banaszczyk et al. (2024)[60] Meta Quest 2; Meta Quest Pro UR5e robot replaying a seated gameplay trajectory captured with OptiTrack, about 0.11-0.13 cubic meters of movement; hand-eye calibration, time alignment and comparison against robot controller poses For the tested replay, the authors judged the difference between headsets to be within calibration accuracy; this was not a universal ranking
Singh, Sharma, Liaqat and Kalawsky (2025)[61] Magic Leap 2; HTC Vive XR Elite Indoor 6 by 4 m factory-like lab, eight Vicon Vero cameras; ML2 OS 1.9.0, Vive XR version 7.3/System 1.0.999.702, Unity 2022.3 LTS; coordinate alignment and fixed-offset correction, poses averaged over 10 frames Mean positional error 68.188 mm (standard deviation 37.911 mm) for Magic Leap 2 and 124.510 mm (standard deviation 61.212 mm) for Vive XR Elite; reported mean anchor drift 0.6 mm and 30.1 mm in a separate measuring-tape test after about 2-3 min of movement

Niehorster and colleagues concluded that the precision of the first-generation Vive was high and its latency low, but that the shifting offset between the virtual and physical tracking space made it unsuitable at the time for experiments that need accurate visual simulation of self-motion.[39] Holzwarth and colleagues judged the Quest 2 suitable for a wide range of research and industrial uses, while noting that latency and rooms with fewer landmarks or different lighting still needed study.[59]

See also

References

  1. ↑ 1.0 1.1 1.2 "Positional Tracking Overview". StereoLabs Documentation. StereoLabs. https://docs.stereolabs.com/docs/development/zed-sdk/modules/positional-tracking. Retrieved 2026-10-12.
  2. ↑ 2.00 2.01 2.02 2.03 2.04 2.05 2.06 2.07 2.08 2.09 2.10 2.11 2.12 2.13 2.14 2.15 Steven M. LaValle (2019). "Virtual Reality, Chapter 9: Tracking". University of Oulu. https://msl.cs.uiuc.edu/vr/vrch9.pdf. Retrieved 2026-10-12.
  3. ↑ 3.0 3.1 3.2 3.3 3.4 3.5 Yuval Boger (2014-06-02). "Overview of Positional Tracking Technologies for Virtual Reality". Road to VR. https://www.roadtovr.com/overview-of-positional-tracking-technologies-virtual-reality/. Retrieved 2026-10-12.
  4. ↑ 4.0 4.1 4.2 4.3 4.4 4.5 4.6 Diederick C. Niehorster, Li Li, Markus Lappe (2017-05-18). "The Accuracy and Precision of Position and Orientation Tracking in the HTC Vive Virtual Reality System for Scientific Research". i-Perception 8 (3). doi:10.1177/2041669517708205. Retrieved 2026-10-12.
  5. ↑ 5.0 5.1 5.2 5.3 "Rigid Body Tracking". OptiTrack Documentation, Motive 3.0. OptiTrack. https://docs.optitrack.com/v3.0/motive/rigid-body-tracking. Retrieved 2026-10-12.
  6. ↑ 6.0 6.1 6.2 6.3 6.4 "The OpenXR Specification, Chapter 7: Spaces". Khronos OpenXR Registry. Khronos Group. https://registry.khronos.org/OpenXR/specs/1.1/html/xrspec.html#spaces. Retrieved 2026-10-12.
  7. ↑ 7.0 7.1 7.2 7.3 7.4 Christian Forster, Luca Carlone, Frank Dellaert, Davide Scaramuzza (2017-02). "On-Manifold Preintegration for Real-Time Visual-Inertial Odometry". IEEE Transactions on Robotics 33 (1): 1-21. doi:10.1109/TRO.2016.2597321. Retrieved 2026-10-12.
  8. ↑ 8.0 8.1 8.2 8.3 8.4 8.5 8.6 8.7 Greg Welch, Eric Foxlin (2002). "Motion Tracking: No Silver Bullet, but a Respectable Arsenal". IEEE Computer Graphics and Applications 22 (6): 24-38. doi:10.1109/MCG.2002.1046626. Retrieved 2026-10-12.
  9. ↑ "IS-900 System". InterSense. Thales. https://www.intersense.com/is-900-system/. Retrieved 2026-10-12.
  10. ↑ 10.0 10.1 10.2 Wenda Zhao, Abhishek Goudar, Angela P. Schoellig (2022). "Finding the Right Place: Sensor Placement for UWB Time Difference of Arrival Localization in Cluttered Indoor Environments". arXiv:2204.04508. https://arxiv.org/pdf/2204.04508. Retrieved 2026-10-12.
  11. ↑ "Sub Millimeter 3D Wireless Positional Tracking for Location Based VR (LBVR)". IndoTraq. IndoTraq LLC. 2018-09. https://www.indotraq.com/wp-content/uploads/Whitepaper-LBVR.pdf. Retrieved 2026-10-12.
  12. ↑ "Our Technology". Polhemus. https://polhemus.com/technology. Retrieved 2026-10-12.
  13. ↑ "Electromagnetic Motion Tracking and Metallic Distortion: Is it a Problem and Is There a Solution?". Polhemus. https://polhemus.com/blog/polhemus-introduces-ftt. Retrieved 2026-10-12.
  14. ↑ "Thanks To The Razer Hydra, Now You’re Thinking With Motion Control". Razer Newsroom. 2011-04-21. https://www.razer.com/newsroom/product-news/thanks-to-the-razer-hydra-now-youre-thinking-with-motion-control. Retrieved 2026-10-12.
  15. ↑ 15.0 15.1 "SteamVR Tracking". Steamworks. Valve Corporation. https://partner.steamgames.com/vrlicensing. Retrieved 2026-10-12.
  16. ↑ "Detection of ArUco Markers". OpenCV Documentation, 4.x. https://docs.opencv.org/4.x/d5/dae/tutorial_aruco_detection.html. Retrieved 2026-10-12.
  17. ↑ 17.0 17.1 17.2 17.3 "Fundamental concepts". Google for Developers. Google. https://developers.google.com/ar/develop/fundamentals. Retrieved 2026-10-12.
  18. ↑ 18.0 18.1 Joel Hesch, Anna Kozminski, Oskar Linde (2019-08-22). "Powered by AI: Oculus Insight". Meta AI. https://ai.meta.com/blog/powered-by-ai-oculus-insight/. Retrieved 2026-10-12.
  19. ↑ 19.0 19.1 Jamie Shotton, Andrew Fitzgibbon, Mat Cook, Toby Sharp, Mark Finocchio, Richard Moore, Alex Kipman, Andrew Blake (2011). "Real-Time Human Pose Recognition in Parts from Single Depth Images". Microsoft Research. https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/BodyPartRecognition.pdf. Retrieved 2026-10-12.
  20. ↑ Richard A. Newcombe, Shahram Izadi, Otmar Hilliges, David Molyneaux, David Kim, Andrew J. Davison, Pushmeet Kohli, Jamie Shotton, Steve Hodges, Andrew Fitzgibbon (2011). "KinectFusion: Real-Time Dense Surface Mapping and Tracking". Proceedings of ISMAR 2011. https://www.microsoft.com/en-us/research/wp-content/uploads/2016/11/ismar_2011.pdf. Retrieved 2026-10-12.
  21. ↑ Nirav Patel (2017-10-10). "Open Source Release of Rift DK2". Meta Horizon Developer Blog. https://developers.meta.com/horizon/blog/open-source-release-of-rift-dk2/?locale=es_ES. Retrieved 2026-10-12.
  22. ↑ 22.0 22.1 Andrew Melim (2019-09-20). "Increasing Fidelity with Constellation-Tracked Controllers". Meta Horizon Developer Blog. https://developers.meta.com/vr/blog/increasing-fidelity-with-constellation-tracked-controllers/. Retrieved 2026-10-12.
  23. ↑ 23.0 23.1 23.2 "Understanding World Tracking". Apple Developer Documentation. https://developer.apple.com/documentation/arkit/understanding-world-tracking. Retrieved 2026-10-12.
  24. ↑ 24.0 24.1 Santiago Cortés, Arno Solin, Esa Rahtu, Juho Kannala (2018-07-25). "ADVIO: An Authentic Dataset for Visual-Inertial Odometry". arXiv:1807.09828. https://arxiv.org/pdf/1807.09828. Retrieved 2026-10-12.
  25. ↑ "PlayStation VR2 tech specs". PlayStation.com. Sony Interactive Entertainment. https://www.playstation.com/en-us/ps-vr2/ps-vr2-tech-specs/. Retrieved 2026-10-12.
  26. ↑ "ROS 2 - ZED Stereo Node". StereoLabs Documentation. https://docs.stereolabs.com/docs/integrations/ros-2/zed-stereo-node. Retrieved 2026-10-12.
  27. ↑ 27.0 27.1 "Managing Session Life Cycle and Tracking Quality". Apple Developer Documentation. https://developer.apple.com/documentation/arkit/managing-session-life-cycle-and-tracking-quality. Retrieved 2026-10-12.
  28. ↑ "WorldSense". Google VR Documentation. Google. https://developers.google.com/vr/discover/worldsense. Retrieved 2026-10-12.
  29. ↑ "Motion controllers". Microsoft Learn. Microsoft. 2022-05-25. https://learn.microsoft.com/en-us/windows/mixed-reality/design/motion-controllers. Retrieved 2026-10-12.
  30. ↑ 30.0 30.1 30.2 30.3 30.4 30.5 30.6 30.7 Greg Welch, Eric Foxlin (2002). "Motion Tracking: No Silver Bullet, but a Respectable Arsenal". IEEE Computer Graphics and Applications, vol. 22, no. 6. pp. 24-38. https://doi.org/10.1109/MCG.2002.1046626. Retrieved 2026-10-12.
  31. ↑ 31.0 31.1 31.2 31.3 31.4 Johnny Chung Lee. "Johnny Chung Lee - Projects - Wii". johnnylee.net. http://johnnylee.net/projects/wii/. Retrieved 2026-10-12.
  32. ↑ 32.0 32.1 32.2 "PlayStation Move Motion Controller Delivers a Whole New Entertainment Experience to PlayStation 3". Sony Interactive Entertainment. Sony Computer Entertainment. 2010-03-10. https://sonyinteractive.com/en/press-releases/2010/playstationmove-motion-controller-delivers-a-whole-new-entertainment-experience-to-playstation3/. Retrieved 2026-10-12.
  33. ↑ 33.0 33.1 33.2 Ben Lang (2014-03-19). "GDC 2014: Oculus Rift Developer Kit 2 (DK2) Pre-orders Start Today for $350, Ships in July". Road to VR. https://roadtovr.com/oculus-rift-developer-kit-2-dk2-pre-order-release-date-specs-gdc-2014/. Retrieved 2026-10-12.
  34. ↑ 34.0 34.1 34.2 Sid Shuman (2023-02-06). "PlayStation VR2: The ultimate FAQ". PlayStation.Blog. Sony Interactive Entertainment. https://blog.playstation.com/2023/02/06/playstation-vr2-the-ultimate-faq/. Retrieved 2026-10-12.
  35. ↑ 35.0 35.1 35.2 35.3 Nate Kozak (2016-10-14). "Oculus Touch Will Support 360 and Room-scale Tracking With Extra Cameras". Road to VR. https://www.roadtovr.com/oculus-touch-support-room-scale-360-tracking-extra-cameras-sensor/. Retrieved 2026-10-12.
  36. ↑ 36.0 36.1 36.2 Ben Lang (2016-05-03). "One Year Later, Oculus and Valve Still Mum on Timeline to Open Tracking to Third-parties". Road to VR. https://www.roadtovr.com/oculus-constellation-valve-lighthouse-open-tracking-third-party-api/. Retrieved 2026-10-12.
  37. ↑ 37.0 37.1 37.2 "Use Case: Virtual Reality". IndoTraq. IndoTraq LLC. https://indotraq.com/use-case-virtual-reality-and-augmented-reality/. Retrieved 2026-10-12.
  38. ↑ 38.0 38.1 38.2 38.3 38.4 "Sub Millimeter 3D Wireless Positional Tracking for Location Based VR (LBVR)". IndoTraq. IndoTraq LLC. 2018-09. https://www.indotraq.com/wp-content/uploads/Whitepaper-LBVR.pdf. Retrieved 2026-10-12.
  39. ↑ 39.00 39.01 39.02 39.03 39.04 39.05 39.06 39.07 39.08 39.09 Diederick C. Niehorster, Li Li, Markus Lappe (2017). "The Accuracy and Precision of Position and Orientation Tracking in the HTC Vive Virtual Reality System for Scientific Research". i-Perception, vol. 8, no. 3. https://doi.org/10.1177/2041669517708205. Retrieved 2026-10-12.
  40. ↑ 40.0 40.1 40.2 "HoloLens (1st gen) hardware". Microsoft Learn. Microsoft. https://learn.microsoft.com/en-us/previous-versions/mixed-reality/hololens-1/hololens1-hardware. Retrieved 2026-10-12.
  41. ↑ 41.0 41.1 41.2 "How to Check Functionality of the Sensor Bar (Wii)". Nintendo Support. Nintendo. https://en-americas-support.nintendo.com/app/answers/detail/a_id/2954/~/how-to-check-functionality-of-the-sensor-bar-(wii). Retrieved 2026-10-12.
  42. ↑ 42.0 42.1 42.2 42.3 42.4 "PlayStation VR Instruction Manual (CUH-ZVR1)". PlayStation.com. Sony Interactive Entertainment. https://www.playstation.com/content/dam/global_pdc/en/corporate/support/manuals/psvr-docs/cuh-zvr1/EN_PSVR_CUH-ZVR1_Instruction_Manual_Web.pdf. Retrieved 2026-10-12.
  43. ↑ 43.0 43.1 Shuhei Yoshida (2015-03-03). "Project Morpheus: PS4 VR Upgraded, Coming in 2016". PlayStation.Blog. Sony Computer Entertainment. https://blog.playstation.com/2015/03/03/project-morpheus-ps4-vr-upgraded-coming-in-2016/. Retrieved 2026-10-12.
  44. ↑ 44.0 44.1 Ben Lang (2018-05-04). "Lenovo Mirage Solo Review: Positional Tracking Comes to Mobile VR (sort of)". Road to VR. https://roadtovr.com/lenovo-mirage-solo-review-positional-tracking-comes-to-mobile-vr-sort-of/. Retrieved 2026-10-12.
  45. ↑ 45.0 45.1 "WorldSense". Google for Developers. Google. https://developers.google.com/vr/discover/worldsense. Retrieved 2026-10-12.
  46. ↑ 46.0 46.1 46.2 46.3 46.4 46.5 46.6 46.7 Joel Hesch, Anna Kozminski, Oskar Linde (2019-08-22). "Powered by AI: Oculus Insight". Meta AI. Meta. https://ai.meta.com/blog/powered-by-ai-oculus-insight/. Retrieved 2026-10-12.
  47. ↑ 47.0 47.1 47.2 47.3 47.4 "Motion controllers". Microsoft Learn. Microsoft. https://learn.microsoft.com/en-us/windows/mixed-reality/design/motion-controllers. Retrieved 2026-10-12.
  48. ↑ 48.0 48.1 48.2 48.3 48.4 "PS VR2 Tech Specs". PlayStation.com. Sony Interactive Entertainment. https://www.playstation.com/en-us/ps-vr2/ps-vr2-tech-specs/. Retrieved 2026-10-12.
  49. ↑ Oculus VR (2013-01-04). "Building a Sensor for Low Latency VR". Meta Quest Blog. Meta. https://www.meta.com/blog/building-a-sensor-for-low-latency-vr/. Retrieved 2026-10-12.
  50. ↑ Scott Hayden (2018-06-21). "Magic Leap One Controller Appears in FCC Filing, Release on Track for 2018". Road to VR. https://www.roadtovr.com/magic-leap-one-motion-controller-appears-fcc-filing-suggesting-2018-headset-launch/. Retrieved 2026-10-12.
  51. ↑ "IndoTraq LLC Introduces the Fastest and Most Precise Wireless Indoor Tracking System in the World This Week at CES 2016". PRWeb. IndoTraq LLC. 2016-01-06. http://www.prweb.com/releases/2015/12/prweb13145665.htm. Retrieved 2026-10-12.
  52. ↑ Nirav Patel (2017-10-10). "Open Source Release of Rift DK2". Meta Horizon Developer Blog. https://developers.meta.com/horizon/blog/open-source-release-of-rift-dk2/?locale=es_ES. Retrieved 2026-10-12.
  53. ↑ "IndoTraq's High Speed Vision Tracking (HSVT)". IndoTraq. IndoTraq LLC. https://indotraq.com/indotraqs-high-speed-vision-tracking-hsvt/. Retrieved 2026-10-12.
  54. ↑ "SteamVR Tracking". Steamworks. Valve Corporation. https://partner.steamgames.com/vrlicensing. Retrieved 2026-10-12.
  55. ↑ "Base Stations - Valve Index". Valve Index. Valve Corporation. https://www.valvesoftware.com/en/index/base-stations. Retrieved 2026-10-12.
  56. ↑ Ian Hamilton (2019-08-22). "Oculus Insight: Facebook Details Quest's Inside Out Tracking System". UploadVR. https://www.uploadvr.com/oculus-insight-details-quest/. Retrieved 2026-10-12.
  57. ↑ David Heaney (2024-10-01). "Windows Mixed Reality Headsets No Longer Function With Latest Windows 11 Update". UploadVR. https://www.uploadvr.com/windows-11-24h2-kills-windows-mr-support/. Retrieved 2026-10-12.
  58. ↑ Hideaki Nishino (2022-01-04). "PlayStation VR2 and PlayStation VR2 Sense controller: the next generation of VR gaming on PS5". PlayStation.Blog. Sony Interactive Entertainment. https://blog.playstation.com/2022/01/04/playstation-vr2-and-playstation-vr2-sense-controller-the-next-generation-of-vr-gaming-on-ps5/. Retrieved 2026-10-12.
  59. ↑ 59.0 59.1 Valentin Holzwarth, Joy Gisler, Christian Hirt, Andreas Kunz (2021). "Comparing the Accuracy and Precision of SteamVR Tracking 2.0 and Oculus Quest 2 in a Room Scale Setup". Proceedings of the 2021 5th International Conference on Virtual and Augmented Reality Simulations (ICVARS 2021). pp. 42-46. https://doi.org/10.1145/3463914.3463921. Retrieved 2026-10-12.
  60. ↑ Adam Banaszczyk, Mikołaj Łysakowski, Michał R. Nowicki, Piotr Skrzypczyński, Sławomir K. Tadeja (2024). "How Accurate is the Positioning in VR? Using Motion Capture and Robotics to Compare Positioning Capabilities of Popular VR Headsets". 2024 IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct), arXiv preprint. doi:10.1109/ISMAR-Adjunct64951.2024.00027. https://arxiv.org/abs/2412.06116. Retrieved 2026-10-12.
  61. ↑ Shubham Singh, Yash Sharma, Amer Liaqat, Roy S. Kalawsky (2025-07-22). "Evaluation of XR device's real-world tracking accuracy and depth perception from an industrial point of view". Virtual Reality, vol. 29, article 118. https://doi.org/10.1007/s10055-025-01192-3. Retrieved 2026-10-12.

References

Retrieved from "https://vrarwiki.com/index.php?title=Positional_tracking&oldid=49969"
Contents