Experimental Analysis of Radar Odometry by Commercial Ultralight Radar Sensor for Miniaturized UAS

Abstract

Autonomous navigation of miniaturized Unmanned Aircraft Systems (UAS) in complex environments, when Global Positioning System is unreliable or not available, is still an open issue. This paper contributes to that topic exploring the use of radar-only odometry by existing commercial ultralight radars. The focus is set on an end-to-end Multiple-Target Tracking strategy compliant with desired sensor and platform, which exploits both range and bearing measurements provided by the radar. A two-dimensional odometry approach is then implemented. Main results show real-time capabilities and standard deviation of errors in Forward and Cross-range directions smaller than 1.50 m and 3.00 m, respectively. Field test data are also used to discuss the potential of this technique, challenging issues, and future improvement

Ultralight Radar for Small and Micro-Uav Navigation

“…Given a model of radar, the main steps before using odometric algorithms are target detection and multiple-target tracking (MTT) (Scannapieco et al, 2017). Concerning this, it is also necessary to mention that an advantage of radar-based odometry compared with vision-based systems is the direct access to range information that prevents from scale drift phenomena.…”

“…range and azimuth angle of targets, is retrieved from observation of the scene via phase interferometry. This mode of operation differs from both 2D LIDAR and 3D LIDAR (Scannapieco et al, 2017) and it is worth highlighting that range and angular information are provided with an extremely different accuracy and the phase interferometry assumes a single target at each range (Scannapieco et al, 2017).…”

“…Since separation between receivers is very small compared with the range resolution, the magnitude component of two signals (1) is averaged non-coherently to achieve partial clutter suppression thus obtaining the value ( , ). Once range and bearing contents are available, the extraction of relevant information and rejection of clutter are demanded to a one-dimensional Constant False Alarm Rate (CFAR) filter (Scannapieco et al, 2017). Ordered Statistics CFAR (OS-CFAR) (Rohling, 1983) performs well under different operative conditions.…”

“…aerospace and automotive sectors, quicken the integration of ultralight radars onboard small and micro UAV (Moses et al, 2011), , in terms of size, weight and power (SWaP). Radar odometry for UAV navigation has been recently proposed, among others, by (Kaufmann et al, 2015), (Quist et al, 2016a), (Quist et al, 2016b), and (Scannapieco et al, 2017). In particular, (Kaufmann et al, 2015) proposed a simulation of twodimensional navigation solution based on data from both sidelooking radar housed on a fixed-wing UAV and IMU.…”

“…Approaches in (Quist et al, 2016a) and (Quist et al, 2016b) are tailored to fixed-wing UAV and data acquired by a high performance Synthetic Aperture Radar (SAR) flying on a Cessna aircraft. An approach towards small and micro UAV, which can move in different directions with different speeds, hover, or even exhibit only attitude rotations, at very low altitude in a GPSchallenging scenario is presented in (Scannapieco et al, 2017). The work is oriented to existing commercial ultralight radars and to environments that can be significantly cluttered, hindering reliable extraction of many strong and stable scatterers.…”