Automatic Multi-Sensor Extrinsic Calibration For Mobile Robots

Abstract

In order to fuse measurements from multiple sensors mounted on a mobile robot, it is needed to express them in a common reference system through their relative spatial transformations. In this paper, we present a method to estimate the full 6DoF extrinsic calibration parameters of multiple heterogeneous sensors (Lidars, Depth and RGB cameras) suitable for automatic execution on a mobile robot. Our method computes the 2D calibration parameters (x, y, yaw) through a motion-based approach, while for the remaining 3 parameters (z, pitch, roll) it requires the observation of the ground plane for a short period of time. What set this proposal apart from others is that: i) all calibration parameters are initialized in closed form, and ii) the scale ambiguity inherent to motion estimation from a monocular camera is explicitly handled, enabling the combination of these sensors and metric ones (Lidars, stereo rigs, etc.) within the same optimization framework. We provide a formal definition of the problem, as well as of the contributed method, for which a C++ implementation has been made publicly available. The suitability of the method has been assessed in simulation an with real data from indoor and outdoor scenarios. Finally, improvements over state-of-the-art motion-based calibration proposals are shown through experimental evaluation.

Online Extrinsic Calibration based on Per-Sensor Ego-Motion Using Dual Quaternions

“…For the special case of indoor robotics and autonomous vehicles with planar motion, not all calibration parameters are observable since the rotation axes of all transformations are parallel [7]. Therefore, the non-observable parameter z as well as the roll and pitch angle are often estimated from a common ground plane [15], [18], [23] and only the remaining parameters x, y, and yaw angle are estimated from per-sensor motion. Since our approach is based on DQs, we can restrict the optimization on planar calibration only, i.e., x, y, and yaw angle, for the cost of only two additional constraints.…”

“…This also includes the scaling for monocular camera motion estimation. If sensors do not provide simultaneous measurements but provide their data in a common time frame, the motion estimations can be interpolated, for example, similar to [23]. Further, all noise, if not stated differently, is assumed to originate from additive white Gaussian processes.…”

Intrinsic Calibration of Depth Cameras for Mobile Robots Using a Radial Laser Scanner

“…with near zero pitch. The extrinsic calibration parameters between the sensors were estimated using the automatic multi-sensor method proposed in [25]. From the two recorded sequences, one was used to perform the depth calibration described in Section 4, while the other one was used for evaluation purposes.…”

Multi-Sensor Depth Fusion Framework for Real-Time 3D Reconstruction

“…Furthermore, the system is re-calibrated every time before the data acquisition as discussed in [19] which ensures that every numerical discrepancy due to wear and tear is captured and the fusion framework uses that information to achieve higher accuracy. Multi-sensor calibration is a crucial research problem which indirectly affects 3D fusion process, however to keep the focus of this research towards 3D reconstruction, readers are encouraged to see [20]- [22]). Similarly, hardware as well as software considerations have been incorporated by Geiger et.…”