Using 3D point clouds in odometry estimation in robotics often requires finding a set of correspondences between points in subsequent scans. While there are established methods for point clouds of sufficient quality, state-of-the-art still struggles when this quality drops. Thus, this paper presents a novel learning-based framework for predicting robust point correspondences between pairs of noisy, sparse and unstructured 3D point clouds from a light-weight, low-power, inexpensive, consumer-grade System-on-Chip (SoC) Frequency Modulated Continuous Wave (FMCW) radar sensor. Our network is based on the transformer architecture which allows leveraging the attention mechanism to discover pairs of points in consecutive scans with the greatest mutual affinity. The proposed network is trained in a self-supervised way using set-based multi-label classification cross-entropy loss, where the ground-truth set of matches is found by solving the Linear Sum Assignment (LSA) optimization problem, which avoids tedious hand annotation of the training data. Additionally, posing the loss calculation as multi-label classification permits supervising on point correspondences directly instead of on odometry error, which is not feasible for sparse and noisy data from the SoC radar we use. We evaluate our method with an open-source state-of-the-art Radar-Inertial Odometry (RIO) framework in real-world Unmanned Aerial Vehicle (UAV) flights and with the widely used public Coloradar dataset. Evaluation shows that the proposed method improves the position estimation accuracy by over 14 % and 19 % on average, respectively. The open source code and datasets can be found here: https://github.com/aau-cns/radar_transformer.

In this paper, we present a Radar-Inertial Odometry (RIO) method based on the nonlinear optimization of factor graphs in a sliding window fashion. Our method makes use of a light-weight, low-power, inexpensive and commonly available hardware enabling easy deployment on small Unmanned Aerial Vehicles (UAV)s. We keep the state estimation problem bounded by employing partial marginalization of the oldest states, rendering the method real-time capable. We compare the implemented approach to the state-of-the-art multi-state Extended Kalman Filter (EKF)-based method in a one-to-one fashion. That is, we implemented in a single custom C++ RIO framework both estimation back-ends with all other parts shared and thus identical for a fair direct comparison. In the real-world flight experiments, we compare the two methods and show that both perform similarly in terms of accuracy when the linearization point is not far from the true state. Upon wrong initialization, the factor graph approach heavily outperforms the EKF approach. We also acknowledge that the influence of undetected outliers can overwhelm the inherent benefits of the nonlinear optimization approach leading to the insight that the estimator front-end has an important (and often underestimated) role in the overall performance. The open source code and datasets can be found here: https://github.com/aau-cns/aaucns_rio.

In this paper, we present a Radar-Inertial Odometry (RIO) approach that utilizes performance improving modules, enhanced for the sparse and noisy radar signals, from the vision community in order to estimate the full 6DoF pose and 3D velocity of a robot in an unprepared environment. Our method leverages a multi-state approach in which we make use of several past robot poses and trails of measurements from a lightweight and inexpensive Frequency Modulated Continuous Wave (FMCW) radar sensor. Furthermore, in our estimation framework we include a method for promoting measurement trails to persistent landmarks which correspond to salient features in the environment. In an Extended Kalman Filter (EKF) framework, we fuse the range measurements to the persistent landmarks, trails, and the velocity measurements of the detected 3D points together with the Inertial Measurement Unit (IMU) readings. Our method is particularly relevant for (but not limited to) Unmanned Aerial Vehicles (UAV), enabling them to localize while performing missions in Global Navigation Satellite System (GNSS)-denied environments and, thanks to the properties of the radar sensor, in environments generally challenging for robot perception due to external factors such as smoke or extreme illumination. We show in real flight experiments the effectiveness of our estimator and compare it to the state-of-the-art.

Multicopter Unmanned Aerial Vehicles (UAV) are small and agile robots with the potential to become prominent in performing autonomous tasks in various Global Navigation Satellite System (GNSS)-denied environments. These environments can potentially be rendered even more challenging due to external factors impairing the robot's perception, such as low or too bright light, permeation with aerosols or smoke. A precondition of autonomous operation, though, is the ability of a robot to accurately localize itself in the surrounding environment. Millimeter-wave Frequency Modulated Continuous Wave (FMCW) radar sensors are resilient to the aforementioned factors while being lightweight, inexpensive and highly accurate. In this paper, we present a Radar-Inertial Odometry (RIO) method for estimating the full 6DoF pose and 3D velocity of a UAV. In an Extended Kalman Filter (EKF) framework, we fuse range measurements and velocity measurements of 3D points detected by an FMCW radar sensor together with Inertial Measurement Unit (IMU) readings. In real experiments we show that our approach enables accurate state estimation of a UAV and that it exhibits improvements over similar existing state-of-the-art method.