Determining the state of a mobile robot is an essential building block of robot navigation systems. In this letter, we address the problem of estimating the robot's pose in an indoor environment using 2D LiDAR data and investigate how modern environment models can improve gold standard Monte-Carlo localization (MCL) systems. We propose a neural occupancy field to implicitly represent the scene using a neural network. With the pretrained network, we can synthesize 2D LiDAR scans for an arbitrary robot pose through volume rendering. Based on the implicit representation, we can obtain the similarity between a synthesized and actual scan as an observation model and integrate it into an MCL system to perform accurate localization. We evaluate our approach on self-recorded datasets and three publicly available ones. We show that we can accurately and efficiently localize a robot using our approach surpassing the localization performance of state-of-the-art methods. The experiments suggest that the presented implicit representation is able to predict more accurate 2D LiDAR scans leading to an improved observation model for our particle filter-based localization.
30 June 2023 in IEEE Robotics and Automation Letters
Mapping an environment is essential for several robotic tasks, particularly for localization. In this letter, we address the problem of mapping the environment using LiDAR point clouds with the goal to obtain a map representation that is well suited for robot localization. To this end, we utilize a neural network to learn a discretization-free distance field of a given scene for localization. In contrast to prior approaches, we directly work on the sensor data and do not assume a perfect model of the environment or rely on normals. Inspired by the recently proposed NeRF representations, we supervise the network by points sampled along the measured beams, and our loss is designed to learn a valid distance field. Additionally, we show how to perform scan registration and global localization directly within the neural distance field. We illustrate the capabilities to globally localize within an indoor environment utilizing a particle filter as well as to perform scan registration by tracking the pose of a car based on matching LiDAR scans to the neural distance field.
Remote sensing and robotics often rely on visual odometry (VO) for localization. Many standard approaches for VO use feature detection. However, these methods will meet challenges if the environments are feature-deprived or highly repetitive. Fourier-Mellin Transform (FMT) is an alternative VO approach that has been shown to show superior performance in these scenarios and is often used in remote sensing. One limitation of FMT is that it requires an environment that is equidistant to the camera, i.e., single-depth. To extend the applications of FMT to multi-depth environments, this paper presents the extended Fourier-Mellin Transform (eFMT), which maintains the advantages of FMT with respect to feature-deprived scenarios. To show the robustness and accuracy of eFMT, we implement an eFMT-based visual odometry framework and test it in toy examples and a large-scale drone dataset. All these experiments are performed on data collected in challenging scenarios, such as, trees, wooden boards and featureless roofs. The results show that eFMT performs better than FMT in the multi-depth settings. Moreover, eFMT also outperforms state-of-the-art VO algorithms, such as ORB-SLAM3, SVO and DSO, in our experiments.
9 August 2021 in Journal of Intelligent & Robotic Systems