US 12,584,991 B2Granted

UWB-based in-vehicle 3D localization of mobile devices

AssigneeGM Global Technology Operations LLC

InventorsZijun Han, Jinzhu Chen, Aaron Adler, Fan Bai, John Sergakis

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method for in-vehicle localization of a mobile device includes receiving, in real time, sensor data from a plurality of UWB sensors inside a vehicle. The plurality of UWB sensors includes a plurality of UWB anchors and a UWB tag, which is part of the mobile device. The method further includes determining a plurality of location candidates of the UWB tag based on the sensor data received, determining a plurality of sensed current locations of the UWB tag and a plurality of probabilities for each of the plurality of sensed current locations of the UWB tag using a Gaussian Kernel Density Estimation (KDE), tracking a motion of the UWB tag, and determining a real-time position of the UWB tag using a Bayesian estimation.

Claims15

Dependent claims

A method for in-vehicle localization of a mobile device, comprising:

receiving, in real time, sensor data from a plurality of ultra-wideband (UWB) sensors inside a vehicle, wherein the plurality of UWB sensors includes a plurality of UWB anchors each disposed in a fixed position inside the vehicle and at least one UWB tag that is not in a fixed position relative to the vehicle, the UWB tag is part of the mobile device, and the sensor data includes a location information about the UWB anchors and a distance from the UWB tag to each of the UWB anchors;
determining a plurality of location candidates of the UWB tag based on the sensor data received from the plurality of UWB sensors;
determining a plurality of sensed current locations of the UWB tag and a plurality of probabilities for each of the plurality of sensed current locations of the UWB tag using a Gaussian Kernel Density Estimation (KDE) based on the location candidates of the UWB tag;
tracking a motion of the UWB tag by generating a probability matrix (pmot) through applying a Gaussian KDE to motion-predicted locations (Lk), wherein the probability matrix represents a distribution of possible tag locations over time;
and
determining a real-time location of the UWB tag by Bayesian estimation that multiplies the probability matrix (pmot) with a probability matrix (pmea) generated from sensed current locations of the UWB tag via Gaussian KDE, and selecting a maximum probability from the resulting distribution within the vehicle,
wherein the motion of the UWB tag is tracked using a following equations:
L
k
=
L
k
-
1
+
δ
⁢
t
*
v
+
0
.
5
*
a
*
δ
⁢
t
2
p
mot
=
KDE
⁡
(
L
k
)
where:
k is a timestamp;
Lk is a current location at timestamp k;
Lk-1 is a previous location at timestamp k−1;
δt is a time span between timestamp k and timestamp k−1
v is a moving speed of the UWB tag;
a is a movement acceleration of the UWB tag:
pmot is a probability matrix predicted by a motion tracking;
and
KDE is a Kernel Density Estimation,
wherein the Bayesian estimation uses a following equation:
α
=
max
⁡
(
p
mot
k
*
p
mea
k
)
,
where:
pmotk is the probability matrix predicted by the motion tracking;
pmeak is a probability matrix predicted by a sensor measurement through the KDE;
and
α is a maximum probability chosen from element-wise-multiplication between pmotk and pmeak.

67 more elements

A system for in-vehicle localization of a mobile device, comprising:

a plurality of ultra-wideband (UWB) sensors inside a vehicle, wherein the plurality of UWB sensors includes a plurality of UWB anchors each disposed in a fixed position inside the vehicle and at least one UWB tag that is not in a fixed position relative to the vehicle, and the UWB tag is part of the mobile device;
a controller in communication with the plurality of UWB sensors, wherein the controller is programmed to:
receiving, in real time, sensor data from the plurality of UWB sensors, wherein the sensor data includes a location information about the UWB anchors and a distance from the UWB tag to each of the UWB anchors;
determine a plurality of location candidates of the UWB tag based on the sensor data received from the plurality of UWB sensors;
determine a plurality of sensed current locations of the UWB tag and a plurality of probabilities for each of the plurality of sensed current locations of the UWB tag using a Gaussian Kernel Density Estimation (KDE) based on the location candidates of the UWB tag;
track a motion of the UWB tag by generating a probability matrix (pmot) through applying a Gaussian KDE to motion-predicted locations (Lk), wherein the probability matrix represents a distribution of possible tag locations over time;
and
determine a real-time location of the UWB tag by Bayesian estimation that multiplies the probability matrix (pmot) with a probability matrix (pmea) generated from sensed current locations of the UWB tag via Gaussian KDE, and selecting a maximum probability from the resulting distribution within the vehicle,
wherein the motion of the UWB tag is tracked using a following equations:
L
k
=
L
k
-
1
+
δ
⁢
t
*
v
+
0
.
5
*
a
*
δ
⁢
t
2
p
mot
=
KDE
⁡
(
L
k
)
where:
k is a timestamp;
Lk is a current location at timestamp k;
Lk-1 is a previous location at timestamp k−1;
δt is a time span between timestamp k and timestamp k−1
v is a moving speed of the UWB tag;
a is a movement acceleration of the UWB tag:
pmot is a probability matrix predicted by a motion tracking;
and
KDE is a Kernel Density Estimation,
wherein the Bayesian estimation uses a following equation:
α
=
max
⁡
(
p
mot
k
*
p
mea
k
)
,
where:
pmotk is the probability matrix predicted by the motion tracking;
pmeak is a probability matrix predicted by a sensor measurement through the KDE;
and
α is a maximum probability chosen from element-wise-multiplication between pmotk and pmeak.

69 more elements

A tangible, non-transitory, machine-readable medium, comprising machine-readable instructions, that when executed by a processor, cause the processor to:

receive, in real time, sensor data from a plurality of ultra-wideband (UWB) sensors inside a vehicle, wherein the plurality of UWB sensors includes a plurality of UWB anchors each disposed in a fixed position inside the vehicle and at least one UWB tag that is not in a fixed position relative to the vehicle, the UWB tag is part of a mobile device, and the sensor data includes a location information about the UWB anchors and a distance from the UWB tag to each of the UWB anchors;
determine a plurality of location candidates of the UWB tag based on the sensor data received from the plurality of UWB sensors;
determine a plurality of sensed current locations of the UWB tag and a plurality of probabilities for each of the plurality of sensed current locations of the UWB tag using a Gaussian Kernel Density Estimation (KDE) based on the location candidates of the UWB tag;
track a motion of the UWB tag by generating a probability matrix (pmot) through applying a Gaussian KDE to motion-predicted locations (Lk), wherein the probability matrix represents a distribution of possible tag locations over time;
and
determine a real-time position of the UWB tag using a Bayesian estimation, the plurality of sensed current locations of the UWB tag, the motion of the UWB tag, and the plurality of probabilities for each of the plurality of sensed current locations of the UWB tag;
determining a real-time location of the UWB tag by Bayesian estimation that multiplies the probability matrix (pmot) with a probability matrix (pmea) generated from sensed current locations of the UWB tag via Gaussian KDE, and selecting a maximum probability from the resulting distribution within the vehicle,
wherein the motion of the UWB tag is tracked using a following equations:
L
k
=
L
k
-
1
+
δ
⁢
t
*
v
+
0
.
5
*
a
*
δ
⁢
t
2
p
mot
=
KDE
⁡
(
L
k
)
where:
k is a timestamp;
Lk is a current location at timestamp k;
Lk-1 is a previous location at timestamp k−1;
δt is a time span between timestamp k and timestamp k−1
v is a moving speed of the UWB tag:
a is a movement acceleration of the UWB tag;
pmot is a probability matrix predicted by a motion tracking;
and
KDE is a Kernel Density Estimation,
wherein the Bayesian estimation uses a following equation:
α
=
max
⁡
(
p
mot
k
*
p
mea
k
)
,
where:
pmotk is the probability matrix predicted by the motion tracking;
pmeak is a probability matrix predicted by a sensor measurement through the KDE;
and
α is a maximum probability chosen from element-wise-multiplication between pmotk and pmeak.

68 more elements

Dependencies are inferred from the claim text.

Facts

Application: 18/508,579Filed 14 Nov 2023
Published: US 2025/0155541 A115 May 2025
Granted: 24 Mar 2026
Term ends: Not estimatedThe source records don’t support an estimate.

Classification6 codes

G01S 5/0284
Using radio waves: relative positioning
G01S 5/14
Using radio waves: determining absolute distances from a plurality of spaced points of known location
G01S 5/02213
Receivers: receivers arranged in a network for determining the position of a transmitter
G01S 5/0278
Using radio waves: involving statistical or probabilistic considerations