Precision. Speed. Stability. This is what We Do Best.
KineticSense
PRECISION POSITIONING & MOTION DATA TECHNOLOGY
Accurate data leads to accurate decisions. Based on this simple principle, KineticSense redefinesthe standard for trustworthy precision data in sports environments. Fitogether has pushed GNSS-based positioning to its limits, combining RTK-GNSS and IMU sensors to produce centimeter-level 3D motion data. This system integrates hardware, signal processing, algorithms, and data fusion to capture and analyze movement with exceptional accuracy.
Precise Use of All GNSS Systems for Global Coverage
GNSS satellite systems vary by country and region in terms of operator and orbital distribution, resulting in differing satellite availability across the globe. Some regions favor GPS, while others benefit more from systems like Galileo or BeiDou. To address these variations, KineticSense uses a multiband structure that receives L1, L2, and L5 frequencies in parallel, enabling access to all major GNSS systems—GPS, Galileo, GLONASS, and BeiDou. This ensures stable signal reception and precise positioning worldwide.
However, receiving more satellites doesn't always guarantee better accuracy. Multipath errors, signal reflection, and interference can introduce poor-quality signals that degrade precision.
To solve this, KineticSense continuously evaluates the quality of received signals, automatically excludes those with low reliability or high error risk, and selects only trustworthy satellites using Fitogether’s proprietary algorithm to form the optimal satellite set. This signal filtering strategy ensures high accuracy and consistent performance even in complex environments like stadiums or regions with limited satellite visibility.
Robust Signal Reception in Back-Mounted Use
GNSS tracking devices are typically worn on the upper back, making them susceptible to signal blockage or distortion from the head and torso. Player movement, stadium structures, and body water content can also cause reflection and absorption, degrading signal quality. KineticSense addresses these challenges with an antenna optimized for upward and omnidirectional gain, and a finely tuned radiation pattern that ensures stable reception across various postures and movements.
To reduce multipath errors and signal loss caused by close-to-body placement, KineticSense applies a rear-mounted reflector behind the antenna and incorporates filtering and correction algorithms to improve signal purity. It also employs a low-noise amplifier (LNA) to amplify weak GNSS signals with minimal distortion, along with frequency filtering to suppress external interference. Since timing errors directly translate into position errors, KineticSense uses a TCXO reference clock to improve timing accuracy. This maintains precision during extended sessions and ensures reliable signal stability even in stadium environments with complex RF conditions.
As a result, KineticSense ensures signal continuity and positioning accuracy even when placement and interference challenges overlap. This robust design forms the foundation of reliable GNSS tracking in sports, proving its performance across diverse real-world conditions.
Achieving Centimeter-Level Positioning Accuracy with RTK-GNSS
In sports, precise positioning is critical for distance measurement and tactical analysis. However, standard GNSS can produce meter-level errors due to environmental factors, limiting its reliability. RTK (Real-Time Kinematic) corrects these errors using real-time correction data from a base station, achieving centimeter-level accuracy proven across industries. Yet, implementing RTK in wearable devices is extremely challenging. It requires high-precision antennas, stable real-time communication, strong processing performance, and noise-resistant circuits. In dynamic sports conditions where movement is constant and signal dropouts occur frequently, RTK has rarely been applied effectively.
Drawing on years of GNSS optimization, Fitogether engineered a system that meets the signal quality, computational efficiency, and communication stability required for RTK. By integrating the RTK base station into the Live Hub, it enables field-ready RTK use without extra infrastructure. The system has been tested and validated in training grounds and stadiums for years. As a result, Fitogether’s RTK-GNSS wearable achieved the highest accuracy scores ever recorded in FIFA’s EPTS Quality Programme, consistently delivering centimeter-level precision on the field. It represents a major leap in wearable sports tracking and the world’s first practical application of RTK-GNSS in this domain.
From 2D to True 3D Tracking with IMU Sensor Fusion
GNSS or RTK alone struggles to capture complex 3D movement like posture, rotation, and direction. The IMU (Inertial Measurement Unit), consisting of an accelerometer, gyroscope, and magnetometer, detects fine-grained motion such as rotation, directional change, and acceleration at a high update rate of 100Hz. When fused with RTK-GNSS, which provides centimeter-level absolute positioning, IMU data enables complete 3D tracking that reflects real-world movement. IMU maintains continuity when GNSS signals drop, while RTK corrects long-term IMU drift. This synergy ensures accurate position and motion data even between stillness and rapid transitions.
This complementary integration allows seamless and accurate analysis of complex actions like rapid turns and acceleration changes. As a result, KineticSense delivers true 3D tracking that goes beyond traditional GNSS, enabling precise assessment of posture, direction, and speed. This richer motion data supports performance improvement, injury prevention, and tactical decision-making across sports environments.
Related Publications
- A compact RTK-GNSS device for high-precision localization of outdoor mobile robots, Journal of Field Robotics, 2024
- Implementation and Performance Analysis of RTK-GNSS in Wearable Devices for Athletes in Harsh Environments, Electronics Letters, 2025
- The Design of GNSS/IMU Loosely-Coupled Integration Filter for Wearable EPTS of Football Players, Sensors, 2023
- Robust Methods for Estimating the Orientation and Position of IMU and MARG Sensors, IEEE Sensors Letters, 2022
XtendConnect
Long-Range Wireless Data Communication for Sports
In performance analysis, tracking athletes in real-time across large fields is both essential and technically challenging. XtendConnect combines high-power antennas, frequency selection, and a robust packet structure to cover up to 200 meters and transmit data from dozens of players. Without complex infrastructure, it enables consistent data flow and fast decision-making across the training ground or stadium.
Expanding Coverage with High-Performance Antenna Design
In stadiums, wireless signals can be weakened by players, spectators, and structures. XtendConnect solves this by using both high-power directional and omnidirectional antennas. The directional antenna focuses signal strength over long distances, while the omnidirectional antenna spreads coverage evenly to reduce blind spots.
It operates on the 2.4GHz band to ensure longer range and uses hardware design and output control methods that reduce multipath interference. Even in complex environments with many obstacles, it maintains stable signal strength and tracks player movements in real time. As a result, reliable communication is possible at distances up to 200 meters, essential for sports environments.
Reliable Communication through Smart Packet Design
Network disruptions and wireless interference are common in sports. A brief disconnection can lead to lost data. XtendConnect prevents this with a packet structure that includes backup data. If the connection drops, up to one minute of stored data is transmitted as soon as the link is restored—ensuring nothing is lost.
It also simplifies packet headers and uses CRC (Cyclic Redundancy Check) for fast error detection and reliable retransmission. This allows coaches and trainers to receive motion and biometric data every 0.5 seconds (2Hz), significantly faster than conventional systems that update every few seconds. Even with up to 100 devices connected simultaneously, the system delivers smooth and loss-free data streaming.
Automatic Frequency Selection for Interference-Free Communication
With many devices used by players, coaches, and staff, wireless interference becomes a major issue. XtendConnect includes automatic frequency selection, continuously scanning for the best channels and switching to less crowded ones in real time.
This is especially effective during large matches or training sessions with heavy network traffic. Even under load, the system distributes connections efficiently, allowing coaches and trainers to monitor key data—such as position, acceleration, and heart rate—at a stable 2Hz rate. XtendConnect is more than just a stable wireless connection. It’s a smart, sport-optimized system that integrates hardware and software to boost training efficiency and in-game focus.
AirSync
Wireless Data Upload and Seamless Cloud Sync
In sports, how quickly data reaches analysis is a key competitive factor. AirSync is a wireless upload system that syncs data right after training, streamlining management and speeding up analysis. Its parallel structure allows many devices to upload at once, while optimized packets and compression ensure stable syncing for up to 100 players.
Faster Uploads with Wireless Parallel Transfer
Previously, multiple devices had to be connected to a Dock, downloaded to a PC via cable, and then uploaded again to the server. This serial workflow increased transfer time linearly with the number of devices, meaning that as more athletes were involved, the process became significantly slower. Furthermore, a Dock supports only a limited number of devices. Exceeding that limit requires reconnecting and repeating the process, leading to delays and inefficiency.
AirSync uses a wireless parallel upload system centered on the LiveHub device. With no limit on simultaneous connections, dozens of devices can upload training data at once. This boosts upload speed by over four times compared to previous methods and automatically syncs data to the server immediately after each session. For example, 1 hour of 10Hz tracking data (excluding IMU) from 24 players syncs in under a minute. Tagging data from the Live App is uploaded automatically, allowing immediate analysis.
Multi-Channel Upload Technology for Large Squads
When many devices transmit data at the same time after training, wireless interference can reduce upload speed and stability. AirSync addresses this with automatic multi-channel distribution, allowing each device to select the best available path to avoid signal conflict and maintain stable uploads, even in dense environments.
It also uses packet optimization and lossless compression to maximize transfer efficiency and ensure high-reliability uploads without data loss. With this structure, AirSync can support simultaneous uploads from up to 100 players, offering the scalability and stability needed for efficient data management in large-scale training settings.
VisionSync
Turning Video into Measurable Performance
Video and tracking data are rapidly becoming the new standard in sports analysis. While optical tracking is hard to use in training due to processing load and equipment limits, VisionSync uses AI to align wearable data with video, enabling full analysis in under 10 minutes. This gives teams faster, clearer insights and allows real-time tactical feedback.
AI-Based Player Detection and Coordinate Conversion in Video
VisionSync begins analysis using various types of uploaded match footage, such as drone, GoPro, and panoramic videos. It automatically corrects lens distortion and framing-related warping to improve detection and alignment accuracy.
Instead of processing all video frames, the system uses AI-based object detection to select and analyze only key moments. This lightweight approach reduces unnecessary computation while maintaining high accuracy, and runs smoothly even on Mac systems.
Detected player positions are first expressed in pixel coordinates. However, this format is not suitable for precise analysis or comparison with wearable data. VisionSync detects key field landmarks like touchlines and the center circle and applies homography to convert pixel coordinates into accurate real-world field positions. This allows movement data from video to align with GNSS tracking on a shared spatial reference, enabling meaningful and quantifiable analysis.
Temporal and Spatial Alignment of Video and Wearable Data
Video and GNSS-based wearable data are recorded on different time axes. Video relies on frame-based internal clocks, while wearables use UTC-based absolute timestamps. Without precise alignment, even the same play may appear at different times in each dataset. VisionSync solves this by analyzing player movement heatmaps in both datasets and finding the best-matching offset between the video and GNSS heatmaps. This enables highly accurate synchronization of video and wearable data timestamps, all without requiring any external hardware. Time alignment can also be done quickly and reliably, even during intense training sessions or live matches.
The coordinate systems also differ: video uses pixels, while GNSS uses geographic coordinates. VisionSync first projects GNSS data onto a flat pitch-based coordinate system. Then, it converts video coordinates using automatically detected landmarks (touchlines, penalty boxes, center circles, etc.). Homography transformation ensures both datasets share the same x–y reference plane. This precisely matched state becomes the foundation for downstream analysis such as player ID matching and location correction.
Player ID Matching and Position Refinement
Even when timestamps and coordinate systems are fully aligned, two key differences still remain between video and wearable data. Video detects bounding boxes around players but includes no built-in identity, while wearable data carries player IDs but may experience minor positional drift depending on signal reception.
VisionSync resolves this by aligning both datasets based on player position and minimizing discrepancies through automated correction. It compares detected bounding boxes and wearable positions, identifies the best match, and adjusts coordinates to reduce spatial error. As alignment continues, wearable data aligns more closely with video positions, and each bounding box is matched with the correct player ID. Player movements and IDs are visualized together in the video, ensuring clarity and data integrity. This system works reliably even during substitutions and complex tactics, consistently delivering accurate results without manual work.
Related Publications
- Cost-Efficient and Bias-Robust Sports Player Tracking by Integrating GPS and Video, MLSA 2022, 2023