System and method for inter-sensor calibration

Patent metadata

Field	Value
Publication number	US-12517218-B2
Application number	US-202318175958-A
Country	US
Kind code	B2
Filing date	Feb 28, 2023
Priority date	Jul 28, 2020
Publication date	Jan 6, 2026
Grant date	Jan 6, 2026

Abstract

A method includes generating an initial radar track using radar data and initial radar parameters and an initial camera track using image angular position data, initial camera parameters, and the radar range data in combination with calculating correction parameters to be applied to one of the radar data and the image data by comparing positions for the object from the initial radar track and the initial camera track, the first correction parameters being selected so that, when applied to the data from the other of the radar and the camera, to generate a first corrected track, a degree of correspondence between the first corrected track and the track of the other of the radar and camera is higher than a degree of correspondence between the initial radar track and the initial camera track.

First claim

The invention claimed is:

1 . A method, comprising:
capturing first sensor data with a first sensor and second sensor data with a second sensor in a tracking system, the first and second sensor data corresponding to a path of an object; generating an initial first sensor track using the first sensor data and initial first sensor parameters stored to memory; generating an initial second sensor track using the second sensor data and initial second sensor parameters stored to the memory; comparing the initial first sensor track to the initial second sensor track in a common coordinate system; and determining, based on the comparison of the first and second sensor tracks, first and second correction parameters configured so that, when applied to the first sensor data and the second sensor data, respectively, weighted three dimension positional residuals between the initial first sensor track and the initial second sensor track are minimized, wherein the initial first sensor track and the initial second sensor track are weighted based on qualities of the initial first sensor track and the initial second sensor track influencing confidence in an accuracy of the initial first sensor track and the initial second sensor track.

2 . The method of claim 1 , wherein the first correction parameters are selected to minimize the results of a cost function:

E

track
⁢

t

=

∑

sensor
⁢

i

(

∑

sensor
⁢

j

≠

sensor
⁢

i

(

∑

position
⁢

n
t

(



x
i

n
t

-

x
j

n
t



w
ij

n
t

)

)

)

where

x
i

n
t

is an estimate of the nth position of an object in a global coordinate system belonging to track t, detected based on the first sensor and a latest estimate for the first sensor parameters and wherein


·


w
ij

n
t

represents a weights 2-norm of a user-defined weight matrix

W

i
⁢
j

n
t

.

3 . The method of claim 1 , further comprising determining, based on the first and second sensor data, whether the initial first and second tracks represent a common track of an object and, when the initial first and second tracks represent a common track of an object, selecting first and second sensor correction parameters that minimize a cost function

E
=

∑

tracks
⁢

t

w
t

×

E

track
⁢

t

where w t is a weight assigned to each track t.

4 . The method of claim 3 , further comprising setting a weight for each of the first and second initial tracks to a default value of w t =1.

5 . The method of claim 4 , where the weight for each of the first and second initial tracks is set according to an equation w t =e− λΔτ , where Δτ is a time in seconds since a corresponding one of the initial first and second tracks was observed and λ is a decay-rate constant.

6 . The method of claim 5 , wherein the weight for first initial track is adjusted based on a degree of confidence for the first sensor.

7 . The method of claim 3 , wherein E is minimized using a grid search across all of a relevant group of parameters for the first and second sensors and selecting from the group of parameters a subset of the parameters that minimize the cost function.

8 . The method of claim 3 , wherein E is minimized by making, based on prior data, an estimation for a value for each of a relevant group of parameters for the first and second sensors and sampling a set of values surrounding the estimation for each parameter.

9 . The method of claim 1 , wherein the method is run continuously during op

Assignees

• Trackman As

Inventors

• Søndergaard Thomas
• Tuxen Fredrik
• Ungstrup Michael

Classifications

• G01S13/426

  Scanning radar, e.g. 3D radar (G01S13/66 takes precedence) · CPC title

• A63B2220/806

  Video cameras · CPC title

• A63B69/0002

  for baseball · CPC title

• A63B2220/89

  Field sensors, e.g. radar systems · CPC title

• G01S13/867 Primary

  Combination of radar systems with cameras · CPC title