Whenever a sensor is mounted on a robot hand, it is important to know
the relationship between the sensor and the hand. The problem of deter
mining this relationship is referred to as the hand-eye calibration pr
oblem, Hand-eye calibration is important in at least two types of task
s: (1) map sensor centered measurements into the robot workspace frame
and (2) tasks allowing the robot to precisely move the sensor In the
past some solutions were proposed, particularly in the case of the sen
sor being a television camera. With almost no exception, all existing
solutions attempt to solve a homogeneous matrix equation of the form A
X = XB. This article has the following main contributions. First we sh
ow that there al-e two possible formulations of the hand-eye calibrati
on problem. One formulation is the classic one just mentioned. A secon
d formulation takes the form of the following homogeneous matrix equat
ion: MY = M'YB The advantage of the latter formulation is that the ext
rinsic and intrinsic parameters of the camera need not be made explici
t. Indeed, this formulation directly uses the 3x4 perspective matrices
(M and M') associated with two positions of the camel-a with respect
to the calibration frame. Moreover this formulation together with the
classic one covers a wider range of camera-based sensors to be calibra
ted with respect to the robot hand: single scan-line cameras, stereo h
eads, range finders, etc. Second, we develop a common mathematical fra
mework to solve for the hand-eye calibration problem using either of t
he two formulations. We represent rotation by a unit quaternion and pr
esent two methods: (1) a closed-form solution for solving for rotation
using unit quaternions and then solving for translation and (2) a non
linear technique for simultaneously solving for rotation and translati
on. Third, we perform a stability analysis both for our two methods an
d for the linear method developed by Tsai and Lent (1989). This analys
is allows the comparison of the three methods. In light of this compar
ison, the nonlinear optimization method, which solves for rotation and
translation simultaneously, seems to be the most robust one with resp
ect to noise and measurement errors.