Kermorgant, Olivier and Folio, David and Chaumette, François


In this paper we propose a new on-line sensor self-calibration framework. The approach is to consider the sensor/robot interaction that links the sensor signal variations to the robot velocity. By on-line calibration, we mean only the actual measurements are used to perform calibration under the condition that the interaction matrix is analytically known. This allows us to propose a very simple and versatile formulation of sensor parameter calibration. Various sensors can be considered, and calibration from different sensory data may be considered within the same process. Intrinsic and extrinsic parameters estimation are formulated as a non-linear minimization problem the Jacobian of which can be expressed analytically from the sensor model. Simulations and experiments are presented for a camera observing four points, showing good results in the case of separated intrinsic and extrinsic calibration, and illustrating the possible limitations in the caseof simultaneous estimation



  title = {A new sensor self-calibration framework from velocity measurements},
  author = {Kermorgant, Olivier and Folio, David and Chaumette, François},
  booktitle = {IEEE International Conference on Robotics and Automation (ICRA\'2010)},
  year = {2010},
  address = {Anchorage, Alaska},
  month = may,
  pages = {1524--1529},
  doi = {10.1109/ROBOT.2010.5509219},
  hal = {0544787/},
  ieeexplore = {5509219},
  issn = {1050-4729},
  keywords = {extrinsic calibration, intrinsic calibration, nonlinear minimization problem, robot velocity, sensor parameter calibration, sensor self-calibration framework, sensor-robot interaction, velocity measurements, calibration, cameras, minimisation, motion control, nonlinear programming, parameter estimation, robots, sensors, velocity control}