Abstract

At the core of every navigation system, there exists a sensor fusion

algorithm (e.g., Kalman filter, Particle filter, etc) that combines

measurements from several sensors. Fusing proprioceptive (e.g., linear

and rotational velocity) and exteroceptive (e.g., distance and bearing

to a point feature) measurements from different navigation sensors,

requires that these measurements are expressed with respect to a common

frame of reference. For this reason, the 3D transformation between all

the sensors needs to be known precisely. Estimation of these unknown, or

approximately known, transformations, which is called extrinsic

sensor-to-sensor calibration, is the main subject of this talk. In

particular, I will focus on the extrinsic calibration of sensor pairs

that contain one motion sensor, and present our preliminary results on

the IMU-camera calibration and our analysis of the system's

observability properties. I will conclude my talk by discussing some of

the open questions in the context of motion-induced sensor-to-sensor

extrinsic calibration and present our road map to address them in our

future work.