Localization is a problem of ’where we are’. Localization techniques help people understand their surrounding environment based on extracted position information in a geographic reference map. The development of global navigation satellite system (GNSS), light detection and ranging (LiDAR), computer vision (CV), etc., enables us to apply localization techniques for more specific tasks. Autonomous driving or robotics needs a reliable localization technique that both instantly retrieves an accurate positioning result and detailed information of the environment, which is challenging to realize in an urban environment. In our research, we propose a 3D point cloud localization framework in the urban environment to realize localization of terrestrial laser scanning (TLS) static scans in mobile laser scanning (MLS) point clouds. 3D point cloud localization consists of two steps: place recognition which aims to find the location of a TLS point cloud in a big MLS point cloud and pose refinement which accurately aligns TLS and corresponding MLS point clouds. In place recognition, about 100 cylinder-like objects with 2D coordinate information are retrieved per million points in a TLS scan or a MLS scene, taking nearly 15 seconds. 5 TLS static scans in the experiment are all successfully matched to corresponding MLS scenes by the Gaussian mixture model (GMM) probabilistic estimation based on extracted cylinder-like objects. Results indicate that it is possible to realize place recognition by designing an object feature descriptor instead of a fine feature descriptor, even though TLS and MLS point clouds are collected at different times (2016 and 2020). A cylinder-like object feature descriptor saves time, compared with taking more than 100 seconds per million points during point feature extraction. In pose refinement, point clouds of TLS scans and corresponding MLS scenes are resampled and inputted to a neural network which consists of feature extraction block (FEB), correspondence search block (CSB) and pose estimation block (PEB). The output of the neural network is tuned by the global refinement process to realize pose refinement. The accuracy (£r , £t ), expressed as the mean rotation error £r in degrees (deg) and the mean translation error £t in meters (m), is (0.25 deg, 0.88 m) if the ground truth varies from (0, 0, 0) to (10 deg, 10 deg, 30 deg) in rotation and (0, 0, 0) to (30 m, 30 m, 10 m) in translation w.r.t. the (x, y, z) axis. The accuracy of a point-based registration neural network and traditional registration methods point-to-point iterative closest point (ICP), coherent point drift (CPD) is (∼10 deg, ∼10 m), (0.27 deg, 0.95 m) and (0.40 deg, 1.94 m), respectively. The run time of processing a million points for our neural network, a point-based neural network, point-to-point ICP and CPD is ∼5 seconds, ∼3 seconds, ∼50 seconds and >100 seconds, respectively. Results prove that current point-based neural networks cannot work in an urban area, but our neural network achieves a more accurate result than some traditional registration methods. A neural network for pose refinement does not need prior information and is more efficient and more general than traditional registration methods as well. Overall, our research contributes to a general framework of 3D point cloud localization, incorporating both a traditional feature extraction method and a novel neural network. We hope this localization framework can be extended to other scenarios more than the urban environment, as well as other further applications e.g. urban city reconstruction.

Global Navigation Satellite System (GNSS) has been developed in recent several decades, which provides a new technique for timing, positioning and navigation. And GNSS positioning is a basic service to us, both in daily life and scientific research. During high-precise positioning, ambiguity resolution is a key factor that has a huge influence on the accuracy of the result. And the wrong fixing is themain source of lose of accuracy, so many methods of test are proposed to validate the fixing result. We are interested in the performance of solutions that have been labeled as wrong fixing, and want to check if those wrong fixings should be excluded or accepted during estimation.
At firstwe introduce the basic model and challenges of ambiguity fixing, aswell as the widely usedmethod integer least squares and Z-transformation. Some previous research of distribution of fixed solution also enables us to compute the bounds of baseline residual. Then we focus on an example to do some pre-research, and find out the main research question - how to accept wrong fixing during estimation and the impact under different scenarios or estimation methods. We use a simulation-basedmethod to do the research but the
measurements are generated based on the real ephemeris in certain day.
After giving the double difference measurement model based on code and phase with respect to single or dual frequencies, in short baseline scenario, we define some parameters 1-norm, infinity-norm, weighted 2-norm, and good/bad performance of wrong fixings that might be useful for analysis. And some detailed information in chosen epochs are shown with multiple figures and we derive those which are helpful, such as infinity-normratio. Then we develop 4 validation methods, Infinity-normratio detection (RD),Weighted 2-norm detection (WD), 1-norm baseline residual detection (BD1) and Infinity-norm baseline residual detection (BDi), to check if we could recognize wrong fixings with good performance from all wrong fixings.
We also compute some statistics, such as the success rate of detection, the rate of misdiagnose, and the rate of participation to see whether our validation methods are effective or not. The histogram of wrong fixing or correct fixing corresponds the existed research of distribution well. And the time series analysis also proves the reliability of our validation methods, although there exist some errors due to the small sample size of simulations.
We also apply the multi-epoch least squares and Kalman filter to see the influence of fixing success rate, standard deviation of horizontal residuals, residual bounds and performance of wrong fixings. We find that there are obvious improvements on all of them. An extra experiment is designed to see the impact of atmospheric delays and the results show that it is really different from short baseline scenarios and we need to find more proper threshold to make sure our validation methods work.
Finally, we could draw a conclusion that it is possible to find out wrong fixings that could be accepted, but the threshold for each validation method should be adaptive for different scenarios and Kalman filter is very reliable, with which the wrong fixing is always likely to be excluded during the estimation.