Data processing: measuring – calibrating – or testing – Calibration or correction system – Length – distance – or thickness
Reexamination Certificate
1999-04-05
2002-05-21
Hoff, Marc S. (Department: 2857)
Data processing: measuring, calibrating, or testing
Calibration or correction system
Length, distance, or thickness
C702S033000, C702S085000, C702S104000, C702S150000
Reexamination Certificate
active
06393370
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the invention
The present invention is directed to a method for evaluating the quality of distance measurements and of the appertaining measurement sensors that make it possible for an autonomous mobile system to construct a cellularly structured map of its environment.
2. Description of the Related Art
Given employment of autonomous mobile systems (AMS) in unprepared, everyday environments, these environments not being known a prior to the AMS or, respectively, only in part, it is necessary for the AMS to perceive its surroundings with suitable sensors. For example, see Rencken, W. D., Leuthäusser, I., Bauer, R., Feiten, W., Lawitzky, G., Möller, M., Low-Cost Mobile Robots for Complex Non-Production Environments, Proc. 1st IFAC Int. Workshop on Intelligent Autonomous Vehicles, April 1993. An environment model is then constructed from the sensor measurements, preferably in the form of a cellularly structured environment map, that takes the respective inadequacies of the measuring principle or of the sensor mechanism itself such as, for example, systematic measuring errors into consideration See for example Beckerman, M., Oblow, E. M., Treatment of Systematic Errors in the Processing of Wide-Angle Sonar Sensor Data for Robotic Navigation, IEEE Transactions on Robotics and Automation, Vol. 6, No. 2, April 1990. For handling a commanded task, the AMS implements, for example, functions such as obstacle avoidance, navigation and localization dependent on the momentary surroundings. Mismeasurements can jeopardize the success of such operations. First, undetected obstacles can lead to collisions; second, “ghost obstacles” deteriorate the freedom of mobility of the system. The success of the system operation to be implemented is accordingly highly dependent on the condition of the sensors. Errors of the sensor mechanism beyond the limits that are specified and tolerable within the appertaining application must therefore be dependably recognized. The following error recognition demands are made of the sensors thereof on the basis of the desired, typical use environments of autonomous mobile systems:
independence from the surroundings:
no reference or calibration environments should be necessary.
independence from the type of error:
due to the multitude of error types that are generally not known a priori, the approach should optimally cover all error types that can involve a significant deterioration of the environment model.
online capability:
a constant monitoring of the sensors should be possible for safety reasons.
Sensors are mainly secured to the circumference of a mobile robot for reasons of maximizing the range of perception. As a result thereof, however, they are exposed in part to considerable mechanical stresses due to collisions with external objects. In addition to hardware defects of a sensor, considerable deterioration of the imaging properties of an inherently physically functional sensor can occur due to mechanical influences.
Methods for producing cellularly structured environment maps of autonomous mobile systems are known from the Prior Art. For example, cellularly structured maps are employed that use occupied and free probabilities per cell for marking obstacles and for route planning based thereon see the publications Borenstein, J., Koren, Y., The Vector Field Histogram—Fast Obstacle Avoidance for Mobile Robots, IEEE Transactions on Robotics and Automation, Vol. 7, No. 3, 1991 and Borenstein, J., Koren, Y., Real-Time Obstacle Avoidance for Fast Mobile Robots, IEEE Transactions on Systems, Man, and Cybernetics, 19(5), 1989, 1179-1187For example, they are constructed by the incoming measurements M with the assistance of a simplified probabilistic sensor model, refer to the publication Matthies, L., Elfes, A., Probablistic [sic] Estimation Mechanisms and Testrelated [sic] Representations for Sensor Fusion, Proc. of the SPIE—The International Society for Optical Engineering, 1003, 1988, 2-11.
Solutions for error recognition and calibration for sensor mechanisms of autonomous mobile systems (AMS) without employing reference models and calibration members are hitherto unknown.
SUMMARY OF THE INVENTION
An object underlying the invention is therefore comprised in specifying a method that enables the recognition of sensor errors of an autonomous mobile system on the basis of a cellularly structured presentation of its surroundings and, moreover, supports the calibration of measuring sensors in unprepared environments.
This object is achieved by a method for evaluating the measuring quality of a distance-measuring sensor at an autonomous mobile system, including:
a) obstacles in the environs of the system are measured by a plurality of distance-measuring sensors located at the autonomous mobile system and cells of a cellularly structured environment map corresponding in position with the environs are characterized with respect to their occupancy state with obstacles on the basis of the measured results;
b) which sensors have measured a respective cell is noted for this cell identifiable per measuring sensor;
c) the measuring quality of a first measuring sensor is evaluated, at least with respect to a first cell, to see how many other measuring sensors arrive at the same characterization of the occupancy state with respect to the occupancy state of the first cell, whereby the measuring quality of the first sensor is evaluated all the greater the more of the other sensors confirm its characterization.
Developments of the invention provide that the occupancy state of a respective cell is characterized according to an occupied and free probability, whereby the assigning of the respective probabilities is based on how many measuring sensors have measured an obstacle or, respectively, no obstacle there. The measuring quality of a measuring sensor S
j
is evaluated as follows as a probability P for the sensor condition:
P
(
Y
(
S
j
)=
OK|{K
(
C
)}
t
)=
P
(
S
j
|{K}
t
)
P
(
Y
(
S
j
)=
KO|{K
(
C
)}
t
)=
P
(
S
j
|{K}
t
)=1
−P
(
S
j
|{K}
t
)
with the random variables Y and the statusses OK and KO for functional and malfunctioning, whereby this depends on the consistency K of the cells {C} evaluated up to the point in time t and the consistency represents a criterion for the extent to which the characterization of a measuring sensor S
j
coincides with the characterizations of other measuring sensors with respect to the map cell C
i
under consideration. In a preferred embodiment, a consistency measure according to:
P
(
K
i
|S
j
)=
P
(
K
(
C
i
)=
CON|Y
(
S
j
)=
OK
)
P
(
K
i
S
j
)=
P
(
K
(
C
i
)=
CON|Y
(
S
j
)=
KO
)
is assigned for a measuring sensor and the probability for the sensor condition is updated according to:
P
(
S
j
|
{
K
}
i
)
=
P
(
K
i
|
S
j
)
·
P
(
S
j
|
{
K
}
i
-
1
)
(
P
(
K
i
|
S
j
)
·
(
P
(
S
j
|
{
K
}
i
-
1
)
+
P
(
K
i
|
⫬
S
j
)
·
P
(
⫬
S
j
|
{
K
}
i
-
1
)
(
15
)
The probability P(S
j
|{K}
i
), dependent of the occupation state characterized by the measuring sensor as right P
ok
or wrong P
ko
, is at least calculated as:
P
ok
(
S
j
)=
P
(
C
i
|{M
k,k≠j
}
t
, M
j
=OCC
)=
P
(
C
i
|(
M}
t
)
P
ko
(
S
j
)=
P
(
C
i
|{M
k,k≠j
}
t
,M
j
=FREE
)
for a cell characterized as occupied OCC and is at least calculated as:
P
ok
(
S
j
)=
P
(
C
i
|{M
k,k≠j
}
t
,M
j
=FREE
)=
P
(
C
i
|(
M}
t
)
P
ko
(
S
j
)=
P
(
C
i
|{M
k,k≠j
}
t
,M
J
=OCC
)
for a cell characterized as FREE, with M as plurality of implemented measurements.
The consistency measure is determined at least as:
P
(
K
i
|S
j
)=&agr;·
P
ok
(
S
j
)+(1−&agr;)·(1
−P
ok
(
S
j
))
P
(
K
i
Hoff Marc S.
Schiff & Hardin & Waite
Siemens Aktiengesellschaft
Tsai Carol S. W.
LandOfFree
Method for assessing the measuring accuracy of a sensor... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for assessing the measuring accuracy of a sensor..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for assessing the measuring accuracy of a sensor... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2820849