Full-body tracking

Full-body tracking (often abbreviated FBT) is the use of additional sensors worn on the body to capture the position and movement of body parts beyond the head and hands in virtual reality (VR) and other interactive systems. A standard consumer VR setup tracks only three points: the head-mounted display and two hand-held motion controllers. Full-body tracking adds sensors at locations such as the hips, feet, chest, elbows and knees so that a user's legs and torso are represented directly rather than estimated, letting an avatar reproduce sitting, kneeling, dancing and other whole-body motion.

Because most software solves for the full skeleton from a small number of tracked points, full-body tracking is tied to inverse kinematics, which estimates the joints between tracked targets. The same problem of reconstructing a pose from sparse measurements has been studied in academic motion-capture research, where systems infer a complete body pose from as few as six inertial sensors.

Tracking point configurations

The number of tracked points determines how faithfully an avatar follows the wearer. In consumer VR the headset and two controllers already cover three points (head and both hands), so additional trackers are described by how many extra points they add or by the total point count.

The minimum useful configuration in social VR is three extra trackers, on the waist and both feet, which combined with the headset and controllers gives a six-point solve. Software fills in the legs and spine using inverse kinematics. Adding more trackers reduces the amount the software has to guess: trackers on the knees, chest and elbows or shoulders improve fidelity, and platforms such as VRChat support up to eight additional trackers beyond the headset and controllers.

Configuration	Worn trackers (beyond HMD and controllers)	Tracked points (total)	Body parts directly tracked
Three-point	3	6	Head, hands, waist, both feet
Six-point	6	9	Adds chest and both knees (or elbows)
Higher fidelity	up to 8	up to 11	Adds chest, knees, elbows or shoulders

How it works

Full-body tracking systems differ mainly in how each worn sensor determines where it is. Three approaches are in common use, each with different trade-offs for setup, accuracy and resistance to occlusion.

Lighthouse (base-station) tracking

The most accurate consumer approach uses Valve's Lighthouse system, marketed as SteamVR Tracking. One or more base stations sweep the room with infrared laser planes, and photosensors on each tracker time when the sweeps arrive to compute the tracker's position and orientation. This is a form of positional tracking that delivers low latency and sub-millimetre precision, but it requires mounting base stations and a calibrated play area, and a tracker can briefly lose tracking if the user's body blocks the line of sight to every base station. The Vive Tracker and the Tundra Tracker use this method.

Inertial (IMU) tracking

Inertial systems put an inertial measurement unit (IMU), a combination of accelerometer, gyroscope and usually magnetometer, in each tracker. Each unit measures its own orientation and reports it wirelessly to software that combines the readings with the user's body proportions, using the headset as a positional reference. Because nothing has to see the trackers, IMU systems cannot be occluded by clothing, furniture or the user's own body and need no base stations or cameras.

The drawback of inertial tracking is yaw drift: a gyroscope integrates rotation over time, so small errors accumulate and a tracker's idea of which way it points slowly rotates away from reality, requiring periodic resets. Reducing this drift is the main engineering challenge for IMU products, and the academic field of sparse-inertial pose estimation exists to reconstruct plausible full-body motion from these noisy signals.

Camera and depth-sensor tracking

Optical methods infer body pose from images. Early consumer markerless tracking used Microsoft's Kinect depth sensor, whose first version (2010, Windows version 2012) reconstructed a skeleton of about 20 joints from a single depth camera without markers or a worn suit; the underlying per-pixel body-part classification was developed at Microsoft Research Cambridge. A more recent variant moves the cameras onto the trackers themselves: the HTC Vive Ultimate Tracker carries two wide-field-of-view cameras and runs simultaneous localization and mapping so each tracker locates itself by recognising the room, with no base stations and no need to stay in view of the headset.

Hardware

The following devices are commonly used for full-body tracking in consumer VR. Lighthouse trackers depend on SteamVR base stations; inertial and self-tracking units do not.

Device	Method	Per-unit weight	Battery (approx.)	Notes
Vive Tracker 3.0 (2021)	Lighthouse	75 g	7.5 h	Pairs to a per-tracker USB dongle; 1/4 inch screw mount
Tundra Tracker (2021)	Lighthouse	up to 50 g	~7-9 h	One multi-device dongle (SW3/SW5/SW7) pairs several trackers
HTC Vive Ultimate Tracker (2023)	Inside-out cameras (SLAM)	N/A	up to 7 h	One USB-C dongle supports up to five trackers; works with standalone headsets; 200 USD each
SlimeVR	IMU (Wi-Fi)	N/A	10-15 h	Open-source; sets from 219 USD; 5 trackers cover the legs
HaritoraX 2 (2024)	IMU (9-axis)	6-17 g	50-80 h	Drift reduced by per-unit factory sensor tuning; 299 USD
Sony Mocopi (2023)	IMU	8 g	~10 h	Six sensors on head, hips, wrists and ankles; smartphone app; 449 USD in the US

The Vive Tracker 3.0 weighs 75 grams, lasts up to 7.5 hours and connects through a dedicated USB dongle per tracker. The Tundra Tracker weighs no more than 50 grams and uses a single multi-device dongle that can pair three, five or seven units depending on the model, reducing the number of USB ports a multi-tracker setup needs. The HTC Vive Ultimate Tracker, released in the fourth quarter of 2023, sold for 200 USD per tracker plus a 40 USD dongle, with a three-pack and dongle at 600 USD.

SlimeVR, an open-source project whose firmware and software are published on GitHub, uses IMU trackers that report over Wi-Fi to a server running on a PC or phone; its trackers last 10 to 15 hours and a five-tracker set covering the legs starts at 219 USD. Shiftall's HaritoraX 2, announced in late 2024 at 299 USD, uses 9-axis IMUs and reduces rotation drift by tuning each accelerometer and gyroscope before shipment; its chest and hip units run about 80 hours and the 6 gram thigh units about 50 hours. Sony Mocopi, released in Japan in early 2023 and in the United States in 2023 at 449 USD, is a set of six 8-gram IMU sensors worn on the head, hips, wrists and ankles that connect to a smartphone app and can stream motion into VRChat.

Professional optical motion capture systems from companies such as OptiTrack and Vicon also perform full-body capture, surrounding a volume with high-speed infrared cameras that track retroreflective markers on a performer; they offer high precision but are aimed at studios rather than home users.

Role in virtual reality

Full-body tracking is most associated with social VR, where users are represented by avatars and the realism of body language matters. By adding the legs and torso, FBT lets an avatar sit, lie down, kneel, crouch and move its lower body in ways that a head-and-hands setup can only approximate through inverse kinematics. VRChat is the platform most identified with the feature; its full-body community centres on dancing, choreography, exercise and expressive movement, and the platform detects compatible trackers through SteamVR and exposes a calibration step that aligns the avatar's skeleton to the worn trackers.

The hardware foundation for consumer FBT was the original Vive Tracker, which HTC revealed at CES 2017 as a Lighthouse-tracked puck for turning real objects into VR-tracked ones. HTC sold the tracker to developers and released open-source code so that three trackers could be used for full-body positional tracking, which set the three-tracker (waist and feet) pattern that later products followed.

Beyond avatars, the same idea supports VR fitness and dance applications, where capturing leg and hip motion is necessary to evaluate movement, and it feeds into motion capture workflows where VR trackers are an inexpensive alternative to studio rigs for animating characters. The arrival of self-tracking and smartphone-based systems such as the HTC Vive Ultimate Tracker and Sony Mocopi extended full-body tracking to standalone headsets like those in the Meta Quest line, which lack the external base stations that Lighthouse trackers need.

Academic background

Reconstructing a full skeleton from only a few worn sensors is mathematically under-constrained, and it has been an active research topic. The Sparse Inertial Poser system (von Marcard, Rosenhahn, Black and Pons-Moll, 2017) showed that arbitrary human motion could be captured with six IMUs placed on the wrists, lower legs, back and head by fitting a statistical body model to the orientation and acceleration data over multiple frames. Deep Inertial Poser (Huang, Kaufmann, Aksan, Black, Hilliges and Pons-Moll, 2018) used a bi-directional recurrent neural network to reconstruct full-body pose from the same six IMUs in real time, training on IMU data synthesised from motion-capture datasets. Subsequent work has pushed toward even sparser setups: a 2025 method estimates full-body pose from just three IMUs on the head and wrists, the same points a standard VR headset and controllers already provide.