Automatic Spatial Calibration of Near-Field MIMO Radar With Respect to Optical Depth Sensors

TL;DR

This work tackles the problem of estimating the spatial relationship between an imaging MIMO radar and optical RGB-D sensors in the radar's near-field. It introduces a purpose-built calibration target and a three-stage pipeline: sensor-specific target detection, sensor-specific localization, and automatic spatial registration to recover the rigid transform $(R,t)$ (and scale $S$) between sensor frames. The key contributions include a near-field-target design with four styrofoam spheres and five embedded steel balls, an automatic detection/localization framework for both ToF/MVS RGB-D and radar data, and a robust registration procedure with an optional refinement stage. Experiments with ToF and MVS depth sensing demonstrate millimeter-level accuracy and robustness across view angles and distances, highlighting the method's practicality for integrated radar-vision sensing in challenging near-field scenarios.

Abstract

Despite an emerging interest in MIMO radar, the utilization of its complementary strengths in combination with optical depth sensors has so far been limited to far-field applications, due to the challenges that arise from mutual sensor calibration in the near field. In fact, most related approaches in the autonomous industry propose target-based calibration methods using corner reflectors that have proven to be unsuitable for the near field. In contrast, we propose a novel, joint calibration approach for optical RGB-D sensors and MIMO radars that is designed to operate in the radar's near-field range, within decimeters from the sensors. Our pipeline consists of a bespoke calibration target, allowing for automatic target detection and localization, followed by the spatial calibration of the two sensor coordinate systems through target registration. We validate our approach using two different depth sensing technologies from the optical domain. The experiments show the efficiency and accuracy of our calibration for various target displacements, as well as its robustness of our localization in terms of signal ambiguities.

Figures (9)

• Figure 1: Our calibration estimates the relative rotation $R$ and translation $t$ between an optical RGB-D sensor and an imaging MIMO radar incorporating high angular resolution.
• Figure 2: The calibration is divided into sensor-specific parts for target detection and target localization. To acquire the calibration parameters, we register the localized target points from the optical domain (blue) to points of the radar domain (orange).
• Figure 3: Target confidence of a corner reflector captured by a MIMO radar at 2.6m (left) and 0.3m (right) distance.
• Figure 4: The calibration target consists of four styrofoam spheres ( 5), each with a steel ball ( 2.5) embedded at its center; sphere centers form a square of 6 edge length. A fifth steel ball is centered on the styrofoam back plane.
• Figure 5: Our setup consists of an imaging MIMO radar, a Kinect Azure camera (ToF) and five DSLR cameras (MVS).