The positioning of moving objects using trilateration has been in use for many years. There are multiple steps to this process, and in each step, there are multiple methods to choose between to get the best performance. In this project we focused on improving the performance of the existing methods.
First, we focused on the first positioning step, trilateration from distance samples received from multiple sensors at a single time. In this step we compared two main methods: An iterative algorithm based on recursive least square, compared to a non-iterative approximative algorithm. We compared the algorithms in a simulated noisy environment. We discovered that the iterative algorithm allows using less sensors under certain assumptions, and in addition it gives less noisy positionings in most geometries.
In the next step we focused on improving object tracking using a Kalman filter. We worked on improving the resiliency of the filter to receiving outlier noisy samples. We compared the use of a classical Kalman filter, to using a Kalman filter with outlier rejection using Mahalanobis distance and using a machine learning model. We found that it is better to use Mahalanobis distance for small angular velocities, and it is better to use the machine learning model for high angular velocities.