Application of Kalman filter-based indoor localization using ultra-wideband systems
Akilu, Farouk (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025060420517
https://urn.fi/URN:NBN:fi:amk-2025060420517
Tiivistelmä
This thesis investigated the using of Kalman Filter-based noise filtering to improve how accurately indoor localization
can be performed with Ultra-Wideband (UWB) systems. UWB is great for precise measurements, and a 5 × 5 meter
area was set up with four anchors and a moving tag. Early tests showed that UWB struggled with signal noise,
multipath issues, and problems when the line-of-sight was blocked.
To fix these issues, a Kalman Filter algorithm was used on the ESP32-S3 microcontroller to make position estimates
better by filtering the UWB data in real-time. The tests showed that this led to much better accuracy and stability in
positioning, with smoother movement paths and less noise. Tweaking the Kalman Filter’s settings for process and
measurement noise was key to getting good results even though it is a trial-and-error approach.
Despite the fact hardware challenges occurred, like issues with IMU sensor integration, blending UWB data with the
filtered estimates worked well. The findings suggest that using Kalman filtering is a solid, straightforward way to get
reliable indoor tracking, especially for smart homes and IoT devices. For future work, Extended Kalman Filters (EKF)
and adding more sensors are recommended to improve performance in settings where things are constantly chang-
ing
can be performed with Ultra-Wideband (UWB) systems. UWB is great for precise measurements, and a 5 × 5 meter
area was set up with four anchors and a moving tag. Early tests showed that UWB struggled with signal noise,
multipath issues, and problems when the line-of-sight was blocked.
To fix these issues, a Kalman Filter algorithm was used on the ESP32-S3 microcontroller to make position estimates
better by filtering the UWB data in real-time. The tests showed that this led to much better accuracy and stability in
positioning, with smoother movement paths and less noise. Tweaking the Kalman Filter’s settings for process and
measurement noise was key to getting good results even though it is a trial-and-error approach.
Despite the fact hardware challenges occurred, like issues with IMU sensor integration, blending UWB data with the
filtered estimates worked well. The findings suggest that using Kalman filtering is a solid, straightforward way to get
reliable indoor tracking, especially for smart homes and IoT devices. For future work, Extended Kalman Filters (EKF)
and adding more sensors are recommended to improve performance in settings where things are constantly chang-
ing