Granted Patent B2

US 12,651,104 · App. 17/722,115 · Granted Jun 9, 2026

Inventor: Justin Meiners (Murray, UT)

Assignee: PassiveLogic, Inc.

G06F30/27G06T11/23G06T11/60G06F2111/10

Abstract

Claims (228)

1 . A computer-implemented method of aligning a polygon-like representation that comprises a floor plan, the floor plan having inaccuracies, the computer-implemented method comprising:

identifying vertices of the polygon-like representation;

iteratively:

computing a first cost function which measures how well the vertices of the polygon-like representation are aligned;

computing a second cost function which measures how far the vertices of the polygon-like representation have been moved from their original locations;

evaluating a stopping criteria using the first cost function and the second cost function; and

using machine learning to determine new positions of the vertices of the polygon-like representation using the first cost function and the second cost function; and

when the stopping criteria is met, generating a polygon-based model of the floor plan based on the new positions of the vertices of the polygon-like representation to create a modified floor plan;

wherein the machine learning comprises gradient descent optimization, or a Nelder-Mead optimization method;

wherein the polygon-like representation is two-dimensional; and

wherein the second cost function comprises:

g

⁡

(

x

1

,

…

,

x

n

)

=

∑

i

=

1

n



x

i

-

p

i



2

.

2 . The computer-implemented method of claim 1 , wherein the first cost function comprises:

selecting a minimum angle increment;

identifying positions of the vertices; and

determining segments from the positions of the vertices of the polygon-like representation.

3 . The computer-implemented method of claim 2 , where evaluating a stopping criteria using the first cost function and the second cost function comprises a weighting of the first cost function and a weighting of the second cost function reaching a minimum value.

4 . The computer-implemented method of claim 3 , wherein the weighting of the first cost function is 1.0 and the weighting of the second cost function is 0.5.

5 . The computer-implemented method of claim 4 , wherein the polygon-like representation comprise at least one segment that is a curve, and wherein when a segment that is a curve between a first curve measurement and a second curve measurement joins two line segments, then deleting the segment that is a curve and joining the two line segments.

6 . The computer-implemented method of claim 5 , further comprising a second polygon-like representation with a second segment, the second segment adjacent to a first segment in the polygon-like representation and wherein the first cost function further comprises a preference for the first segment and the second segment to be parallel.

7 . A system to correct a polygon-like representation that comprise a floor plan, the floor plan with inaccuracies, using machine learning optimization, the system comprising: a memory; a processor in operable communication with the memory, the processor configured to correct a two-dimensional polygon-like representation that comprise a floor plan, comprising:

identifying vertices of the polygon-like representation;

iteratively:

computing a first cost function which measures how well the vertices of the polygon-like representation are aligned;

computing a second cost function which measuring how far the vertices of the polygon-like representation have been moved from their original locations;

evaluating a stopping criteria using the first cost function and the second cost function; and

using machine learning to determine new positions of the vertices of the polygon-like representation using the first cost function and the second cost function; and

when the stopping criteria is met, generating a polygon-based model of the floor plan based on the new positions of the vertices of the polygon-like representation;

wherein the machine learning comprises gradient descent optimization, or a Nelder-Mead optimization method;

wherein the polygon-like representation is two-dimensional; and

wherein the second cost function comprises:

g

⁡

(

x

1

,

…

,

x

n

)

=

∑

i

=

1

n



x

i

-

p

i



2

.

8 . The system of claim 7 , wherein the first cost function comprises:

selecting a minimum angle increment;

identify positions of the vertices; and

determining segments from the positions of the vertices of the polygon-like representation.

9 . The system of claim 8 , wherein the minimum angle increment is 7.5 degrees.

10 . The system of claim 9 , further comprising a second polygon-like representation with a second segment, the second segment adjacent to a first segment in the polygon-like representation and wherein the first cost function further comprises a preference for the first segment and the second segment to be parallel.

11 . The system of claim 10 , wherein the preference for the first segment and the second segment to be parallel comprises

f

⁡

(

x

1

,

…

,

x

n

)

=

∑

i

=

1

n



arccos

⁡

(

〈

x

i

+

1

-

x

i

,

x

i

-

x

i

-

1

〉

)

-

θ

i



2

.

12 . A computer-readable non-transitory storage medium configured with data and instructions which upon execution by at least one processor cause one or more devices to perform an optimization method of aligning a polygon-like representation that comprises a floor plan with inaccuracies, the optimization method comprising:

identifying vertices of the polygon-like representation;

iteratively:

computing a first cost function which measures how well the vertices of the polygon-like representation are aligned;

computing a second cost function which measuring how far the vertices of the polygon-like representation have been moved from their original locations;

evaluating a stopping criteria using the first cost function and the second cost function; and

using machine learning to determine new positions of the vertices of the polygon-like representation using the first cost function and the second cost function; and

when the stopping criteria is met, generating a polygon-based model of the floor plan based on the new positions of the vertices of the polygon-like representation;

wherein the machine learning comprises gradient descent optimization, or a Nelder-Mead optimization method;

wherein the polygon-like representation is two-dimensional; and

wherein the second cost function comprises:

g

⁡

(

x

1

,

…

,

x

n

)

=

∑

i

=

1

n



x

i

-

p

i



2

.

13 . The computer-readable non-transitory storage medium of claim 12 , further comprising an ingoing segment and an outgoing segment; and wherein when the stopping criteria is met then simplifying vertices by iterating through each vertex and removing those for which an angle between the ingoing segment and the outgoing segment is beneath some threshold.

14 . The computer-readable non-transitory storage medium of claim 13 , wherein the angle between the ingoing segment and the outgoing segment is determined by

θ

=

arccos

⁡

(

〈

x

i

-

x

i

-

1

,

x

i

+

1

-

x

i

)

〉



x

i

-

x

i

-

1



⁢



x

i

+

1

-

x

i



.

15 . The computer-readable non-transitory storage medium of claim 14 wherein the polygon-based model of the floor plan is used to modify equipment location at a location represented by the floor plan.

16 . The computer-readable non-transitory storage medium of claim 12 , further comprising the polygon-based model of the floor plan configured to determine sensor placement at a location represented by the floor plan.

17 . The computer-implemented method of claim 2 , further comprising:

for each two adjacent segments determining a current angle;

for each two adjacent segments determining a multiple of the minimum angle increment closest to the current angle to designate a corresponding target angle; and

determining a total difference of squares between each current angle and each corresponding target angle.

18 . The system of claim 7 , where evaluating a stopping criteria using the first cost function and the second cost function comprises a weighting of the first cost function and a weighting of the second cost function reaching a minimum value.

19 . The system of claim 8 , further comprising:

for each two adjacent segments determining a current angle;

for each two adjacent segments determining a multiple of the minimum angle increment closest to the current angle to designate a corresponding target angle; and

determining a total difference of squares between each current angle and each corresponding target angle.

20 . The method of claim 1 , further comprising using the modified floor plan to determine at least one required building component.

Assignments (2)

Continuity (1)

References Cited (17)