Indoor Localization

Spotlight on: Indoor Localization

Creative technical knowledge becomes more important to stand out in this ever-evolving digital world. We aim to acquire the best talent, even those still graduating. That is why we offer the opportunity for students of different professions and specialisms to graduate with a thesis at Avular. We are proud to have the best talents joining our team. After graduating, some of them stay at Avular, some continue their studies and some go travel the world. Just before the end of their project, we ask them to write an article about their graduation thesis.

This week, the spotlight is on Thomas Wolf. He is graduating from the Control Systems Technology group at Eindhoven University of Technology with his thesis about indoor localization.

Indoor Localization Using a Moving Horizon Estimator for Multi-Rate Sensor Fusion and Outlier Detection.

This project is concerned with the localization of agents, such as ground robots and unmanned aerial vehicles, in indoor environments. For positioning purposes, these agents are typically equipped with a variety of sensors such as wheel encoders, lidars and IMU (Inertial Measurement Units). Since wheel encoders and lidars are impractical on unmanned aerial vehicles, the most common on-board sensor used is the inertial measurement unit, which consists of an accelerometer, gyroscope and a three-axis magnetometer. Although accurate estimates can be achieved using an inertial measurement unit, position estimation remains challenging due to accumulation of errors resulting in drifting estimates. In this sense, positioning systems such as ultra-wideband offer an elegant external indoor localization method to compensate for the drifting estimates.

For accurate state estimation this thesis introduces an optimization-based framework following a moving horizon scheme for multi-rate sensor fusion of which the measurements are populated with outliers. Multi-rate simply means “multiple sampling rates” which is required due to the faster sampling rates of an inertial measurement unit compared to other indoor/outdoor positioning systems. The multi-rate sensor fusion scheme is implemented without using a zero-order hold on the slow sampled sensor measurements. This algorithm is ultimately used to estimate the position of a manoeuvring unmanned aerial vehicle while using a state-of-the-art motion capture system to provide an accurate ground truth. A detailed study on the establishment of the localization algorithm as well as the influence of parameters such as horizon length, covariances and arrival cost are presented and are compared with the current state-of-the-art and a well-known Kalman filter.


Based on these analyses, the performance of the system has been determined. First, the algorithm can successfully filter out outliers, drastically improving the performance of the system when outliers are present. When there are no outliers, the system’s performance is unaltered, meaning the algorithm does not have any negative effect on performance. However, the computation time of the algorithm can be a bottleneck in some cases. In cases where outliers are heavily present, the method definitely improves the system’s performance. This makes this research a first step towards a future where outliers can easily and efficiently be rejected.