cs.ROMar 16, 2026

A Novel Camera-to-Robot Calibration Method for Vision-Based Floor Measurements

Authors: Jan Andre Rudolph, Dennis Haitz, Markus Ulrich

Organizations: Machine Vision Metrology Lab, Institute of Photogrammetry and Remote Sensing, Karlsruhe Institute of Technology, Germany

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

A novel hand-eye calibration method for ground-observing mobile robots is proposed. While cameras on mobile robots are common, they are rarely used for ground-observing measurement tasks. Laser trackers are increasingly used in robotics for precise localization. A referencing plate is designed to combine the two measurement modalities of laser-tracker 3D metrology and camera-based 2D imaging. It incorporates reflector nests for pose acquisition using a laser tracker and a camera calibration target that is observed by the robot-mounted camera. The procedure comprises estimating the plate pose, the plate-camera pose, and the robot pose, followed by computing the robot-camera transformation. Experiments indicate sub-millimeter repeatability.

Figures & tables

Figure 1: Mobile robot Rita (adapted from Naab (2023) ), used in the proposed experiments. It features a differential drive with a front-mounted caster wheel. A spherically mounted retro-reflector is installed on top within a rotatable mount, maintaining the line of sight to a laser tracker. The nadir-mounted camera remains occluded in this image, with only its active illumination apparent.
Figure 2: Retrieval of the orientation of the camera inside the robot frame. The plate-absolute-transformation absHplt{}^{abs}H_{plt} and the plate-camera-transformation pltHcam{}^{plt}H_{cam} , as well as the robot-absolute-transformation absHrob{}^{abs}H_{rob} on the other side, are combined to robot-camera-transformation robHcam{}^{rob}H_{cam} (figure adapted from Ulrich and Hillemann (2024) ).
Figure 3: Stereo measurement of the reflector nest positions in the referencing plate frame. The plate-camera-transformations pltHcam,0{}^{plt}H_{cam,0} and pltHcam,1{}^{plt}H_{cam,1} are used to calculate the reflector nest positions pltPi{}^{plt}P_{i} by triangulation (turquoise arrows) (figure adapted from Ulrich and Hillemann (2024) ).
Figure 4: Retrieval of the absolute orientation of the referencing plate. Absolute laser tracker measurements absPi{}^{abs}P_{i} are matched to the reflector nest positions w.r.t. the referencing plate, yielding the absolute-plate-transformation absHplt{}^{abs}H_{plt} (figure adapted from Ulrich and Hillemann (2024) ).
Figure 5: Referencing plates used in experiments. While they are arbitrarily posed on the ground, a mobile robot Rita can be placed on them. The colored circles mark flush-mounted reflector nests. The dark texture represents the HALCON calibration target with circular marks. The rest of the white area provides space to move the robot for a few decimeters.
Figure 6: Results of repeated absolute measurements of marks on the floor in an experiment. Each plot is normalized to the overall mean position, see Table 1 . (a) and (c) denote the same point as Table 1 does. (b) and (d) denote another mark on the floor. (a) and (b) are based on the in-house-designed wooden referencing plate. (c) and (d) are based on the bespoke glass referencing plate
DirectionMean Position [mm]Distance Metrics [mm]Approach Angle Range [ ∘ ]
 XYMax from MeanMean from MeanCluster Radius 
left130825.8228525.5770.1310.0980.117[89.48, 90.66]
right130826.0548525.0140.1190.0940.109[-90.40, -89.96]
up130825.8538525.4270.1880.1400.163[0.12, 0.36]
down130826.2248525.1610.0540.0420.052[179.53, 179.68]
upleft130825.9058525.4130.1920.1390.150[45.00, 45.06]
Table 1: Quantitative results of repeated absolute measurements of marks on the floor in an experiment. Individual (non-averaged) measured positions of this experiment are plotted in 6(c) . Mean position is the mean of the measured absolute coordinates of the respective cluster. Max. and mean from mean denote the (max. and mean, respectively) distance to this mean position . Cluster diameter is the max. distance between two points in the cluster. In the last row, those metrics are calculated for all directions together.
More in this topic

Hand-Eye Calibration

Hand-eye calibration estimates the rigid transform between a robot and its camera or sensor for accurate perception-guided motion.

See all 13 papers

Explore similar work

CardsList

  1. Laser-Tracker-Assisted Camera-to-Robot Calibration for Mobile Robots

    Sep 22, 2026Jan A. Rudolph, Öykü Kandemir, Markus UlrichMobile RoboticsHand-Eye Calibration

    Institute of Photogrammetry and Remote Sensing (IPF), Karlsruhe Institute of Technology (KIT), Englerstr. 7, 76131 Karlsruhe, Germany

  2. Lollypop: Camera-to-Motion-Capture Calibration Verification with a Reference Target

    Oct 3, 2026Tianyi Liu, Kevin Harris, Mihika Dave +1Camera Calibration

    Meta Redmond, WA, USA

  3. A comparative study on the accuracy & repeatability of mobile robotic platforms for the delivery of precision NDE measurement

    Sep 3, 2026SeyedMohammadAmin Nabi Pour, S. Gareth Pierce, Randika Vithanage +3Robot Trajectory TrackingNon-Destructive Testing