US2026048431A1PendingUtilityA1
Robot-assisted sheet metal bending loading and unloading path planning method based on svsdf
Assignee: UNIV NANJING POSTS & TELECOMMUNICATIONSPriority: Aug 19, 2024Filed: Sep 25, 2025Published: Feb 19, 2026
Est. expiryAug 19, 2044(~18.1 yrs left)· nominal 20-yr term from priority
Inventors:XU FENGYUMIAO YIYANGDING DAWEILI XINGBAOWANG PENGYUJING LIANGHUANG YANSHEN JINGJINCHENG YANYUN
G05B 2219/40421G05B 2219/45234B25J 9/1664G05B 19/18B21D 5/0281B21D 43/105B25J 9/1666B21D 43/021
71
PatentIndex Score
0
Cited by
0
References
0
Claims
Abstract
A robot-assisted sheet metal bending loading and unloading path planning method based on an SVSDF is provided. The method includes the following steps: Step 1, constructing a two-dimensional pixel coordinate system; Step 2, obtaining an initial motion trajectory of loading and unloading a sheet metal part; Step 3, calculating path safety; Step 4, calculating a path cost; Step 5, calculating a fitness value of the initial motion trajectory of loading and unloading the sheet metal part; Step 6, optimizing the path cost and the path safety to obtain an optimal motion trajectory of loading and unloading the sheet metal part; and Step 7, through derivation of a bending loading and unloading motion relationship, transforming a motion trajectory of loading and unloading a workpiece itself into a motion trajectory of each axis of a bending robot.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1 . A robot-assisted sheet metal bending loading and unloading path planning method based on a Swept Volume Signed Distance Field (SVSDF), comprising the following steps:
Step 1, constructing a two-dimensional pixel coordinate system, wherein a two-dimensional map plane is constructed after binarizing side-view images of three-dimensional models of upper and lower dies based on a sheet metal bender, an upper left corner of the two-dimensional map plane is used as an origin, and two adjacent right-angled edges passing through the origin are an x axis and a y axis, respectively, so that the two-dimensional pixel coordinate system is formed; if the sheet metal part is regarded as a moving object, and a bending point of the sheet metal part is regarded as a moving point, a motion path of loading and unloading the sheet metal part comprises several path points, and each path point is denoted as (x, y, θ); wherein (x, y) is a position parameter of the sheet metal part in the two-dimensional pixel coordinate system; and θ is an attitude angle of the sheet metal part; Step 2, obtaining an initial motion trajectory of loading and unloading the sheet metal part, wherein a Rapidly-Exploring Random Tree (RRT)-Connect algorithm is used to initially plan the motion path of loading and unloading the sheet metal part to obtain an initial motion trajectory of loading and unloading the sheet metal part; and the initially planning process is repeated for N pop times to obtain N pop initial motion trajectories of loading and unloading the sheet metal part; Step 3, calculating path safety, wherein all pixel points which are prone to colliding with bending motion of the sheet metal part or have a distance less than a safe distance from an upper die and a lower die of the sheet metal bender are found, which are marked as obstacle points of interest; and according to the obstacle points of interest and the initial motion trajectory of loading and unloading the sheet metal part, and based on the SVSDF, the path safety value of each initial motion trajectory of loading and unloading the sheet metal part is calculated; Step 4, calculating path cost, wherein assuming that any initial motion trajectory of loading and unloading the sheet metal part has N path points, the specific calculation formula of the path cost is:
path
cost
=
∑
i
=
1
N
-
1
Δ
x
i
2
+
Δ
y
i
2
+
β
Δ
θ
i
2
wherein Δx i and Δy i are differences in horizontal and vertical position parameters between two adjacent path points i and i+1; 1≤i≤N−1;
Δθ i is a difference in attitude angles of the sheet metal part between two adjacent path points i and i+1;
β is an attitude angle weight coefficient, which is a set value;
Step 5, calculating COST t , wherein a fitness value COST t of a t-th initial motion trajectory of loading and unloading the sheet metal part is calculated, 1≤t≤N pop ; and the calculation formula of COST t is:
COST
t
=
γ
1
path
cost
+
γ
2
path
safety
wherein γ 1 is a cost weight coefficient, and 1≤γ 1 ≤1.5, which is a set value;
γ 2 is a safety weight coefficient, and 1≤γ 2 ≤1.5, which is a set value; and
Step 6, obtaining an optimal motion trajectory of loading and unloading the sheet metal part, wherein based on a Water Cycle Algorithm (WCA), the minimum value of N pop COST t in Step 5 is continuously iteratively optimized and found out, and the motion trajectory of loading and unloading the sheet metal part corresponding to the minimum COST t is used as the optimal motion trajectory of loading and unloading the sheet metal part.
2 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 1 , wherein in Step 2, the RRT-Connect algorithm is an improved synchronous bias greedy RRT-Connect algorithm; and the improved synchronous bias greedy RRT-Connect algorithm uses an adaptive elliptical region to sample random points when initially planning the motion path of loading and unloading the sheet metal part, and performs adaptive growth of each tree node based on the influence of an obstacle.
3 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 2 , wherein in Step 2, the method of using an adaptive elliptical region to sample each random point comprises the following steps:
Step 2A-1, establishing an adaptive elliptical region, wherein a starting point S start (x start , θ start ) and a goal point S goal (x goal , y goal , θ goal ) of the motion path of loading and unloading the sheet metal part are used as two focal points of the adaptive elliptical region; and the length l from the upper die to the lower die of the bender is used as a length c short of a short axis of the adaptive elliptical region, so as to construct the adaptive elliptical region which adaptively changes with the starting point and the goal point; Step 2A-2, constructing a unit circle, wherein the origin of the two-dimensional pixel coordinate system in Step 1 is used as the center to construct a unit circle with a radius of one pixel; Step 2A-3, stretching a random point, wherein a point (x c , y c , θ rand ) is randomly generated in the unit circle, and the point is stretched into a point (x, y, θ rand ) in the adaptive elliptical region; the calculation formula of x and y is:
[
x
y
]
=
[
c
long
/
2
0
0
c
short
/
2
]
[
x
c
y
c
]
wherein c long is a length of a long axis of the adaptive elliptical region, which is calculated according to c short and a focal length between two focal points; and
Step 2A-4, rotating a random point, wherein the point (x, y, θ rand ) is rotated according to two focuses of the adaptive elliptical region, and the rotated point is a randomly sampled point S rand (x rand , y rand , θ rand ); wherein the calculation formula of x rand and y rand is:
[
x
rand
y
rand
]
=
R
[
x
y
]
+
[
(
x
start
+
x
goal
)
/
2
(
y
start
+
y
goal
)
/
2
]
wherein R is a rotation matrix, and the expression is:
R
=
[
cos
α
-
sin
α
sin
α
cos
α
]
wherein α is a rotation angle of the point (x, y, θ rand ), and the specific calculation formula is:
α
=
arc
tan
(
❘
"\[LeftBracketingBar]"
x
start
-
x
goal
❘
"\[RightBracketingBar]"
/
❘
"\[LeftBracketingBar]"
y
start
-
y
goal
❘
"\[RightBracketingBar]"
)
.
4 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 3 , wherein in Step 2, the adaptive growth method of each search tree node comprises the following steps:
Step 2B-1, calculating an expansion direction v 0 of tree nodes without obstacle influence, wherein the specific calculation formula is:
v
0
=
(
[
x
rand
y
rand
]
-
[
x
nearest
y
nearest
]
)
/
[
x
rand
y
rand
]
-
[
x
nearest
y
nearest
]
2
wherein x nearest and y nearest are position parameters of the generated tree node S nearest closest to the randomly sampled point S rand ;
Step 2B-2, finding the total number N o of obstacles influencing a newly generated tree node S new , wherein the randomly sampled point S rand is used as the center to establish a square with a side length as a set pixel a; and the total number of pixel points intersecting the square with the upper die or the lower die of the bender is denoted as the total number N o of obstacles influencing the newly generated tree node S new (x new , y new , θ new );
Step 2B-3, calculating the influence direction v obs of the obstacle on S new , wherein the specific calculation formula is:
v
obs
=
∑
k
=
1
N
0
(
[
x
rand
y
rand
]
-
[
x
k
y
k
]
)
/
[
x
rand
y
rand
]
-
[
x
k
y
k
]
2
wherein x k and y k are pixel coordinates of the k-th obstacle; 1≤k≤N 0 ;
Step 2B-4, calculating the expansion direction v of search tree nodes with obstacle influence, wherein the specific calculation formula is:
v
=
γ
3
v
0
+
γ
4
v
obs
wherein γ 3 +γ 4 =1, γ 3 and γ 4 the set weight coefficients;
Step 2B-5, calculating x new and y new , wherein S nearest which is closest to S rand is used as the starting point of expansion, and a set equidistant step size r p is used to expand in the v direction; the specific calculation formula of x new and y ne is:
[
x
new
y
new
]
=
[
x
nearest
y
nearest
]
+
r
p
v
Step 2B-6, calculating θ new , wherein θ nearest is used as the initial value, and a set equiangular step size r o is used to expand to θ rand , the specific calculation formula is:
θ
new
=
r
o
(
θ
rand
-
θ
nearest
)
+
θ
nearest
wherein θ nearest is an attitude angle of the sheet metal part of the generated tree node S nearest which is closest to the random sampled point S rand ; and
Step 2B-7, determining whether there is a collision in a tree node expansion process, performing resampling if there is a collision, and adding a new node S new to the search tree if there is no collision.
5 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 1 , wherein in Step 3, the method of calculating path safety specifically comprises the following steps:
Step 3-1, calculating a swept volume, wherein a Shape function is used to calculate the total number of pixel points occupied by any initial motion trajectory of loading and unloading the sheet metal part when moving from the starting point S start to the goal point S goal , which is denoted as the swept volume; Step 3-2, constructing a binary image of the swept volume, wherein the swept volume obtained in Step 3-1 is input into a blank two-dimensional map plane by using a Map function, and the binary image is marked as black, thereby forming the binary image of the swept volume; Step 3-3, constructing an SVSDF matrix, wherein any pixel coordinate in the binary image of the swept volume constructed in Step 3-2 is calculated to the SVSDF value of the swept volume, thereby forming the SVSDF matrix; Step 3-4, finding obstacle points of interest, wherein the upper die and the lower die of the sheet metal bender are obstacles in the bending motion of the sheet metal part, all pixel points which are prone to colliding with bending motion of the sheet metal part or have a distance less than a safe distance from an upper die and a lower die are found from the two-dimensional map plane constructed in Step 1, which are marked as obstacle points of interest, and the pixel coordinate of each obstacle point of interest is recorded; Step 3-5, obtaining an SVSDF value of the obstacle point of interest, wherein according to the pixel coordinate of the obstacle point of interest in Step 3-4, the corresponding SVSDF value is found from the SVSDF matrix constructed in Step 3-3, so as to obtain the SVSDF value of each obstacle point of interest; Step 3-6, calculating an average average_sdf of the SVSDF value of the obstacle point of interest; Step 3-7, using normal distribution to determine the position of average_sdf and a probability density value pdf_value; and Step 3-8, calculating path safety, wherein path safety=K×pdf_value; K is a magnitude adjustment coefficient.
6 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 5 , wherein in Step 3-1, the t-th initial motion trajectory of loading and unloading the sheet metal part comprises N path points, that is, a starting point S start , a second path point S 2 , a third path point S 3 , . . . , an i-th path point S i , a (N−2)-th path point S N−2 , and a goal point S goal ; wherein 2≤i≤N−2; and the pixel point occupied by the sheet metal part at the path point S i is Shape(S i ).
7 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 6 , wherein in Step 3-2, the binary image SV_map of the swept volume of the t-th initial motion trajectory of loading and unloading the sheet metal part is:
SV_map
=
Map
(
⋃
i
∈
[
start
,
goal
]
Shape
(
S
i
)
)
.
8 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 7 , wherein in Step 3-3, the calculation formula of the SVSDF is:
SVSDF
=
d
(
SV_map
)
-
d
¯
(
SV_map
)
wherein d(SV_map) is a distance from any pixel point in the binary image of the swept volume to the nearest pixel point of the swept volume in unit of pixels;
d (SV_map) is a complement of d(SV_map).
9 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 7 , wherein in Step 3-4, before finding the obstacle point of interest, the upper die and the lower die of the sheet metal bender are divided into a narrow region and a non-narrow region; the narrow region refers to the region between the protrusion of the bottom of the upper die and the groove of the lower die; the non-narrow region refers to other bending moving regions except the narrow region; and therefore, the obstacle points of interest comprise obstacle points of interest in the narrow region and obstacle points of interest in the non-narrow regions;
the obstacle points of interest in the narrow region comprise pixel points located at an outer edge and a periphery of the bottom of the upper die in the narrow region, and pixel points located at an outer edge and a periphery of the top of the lower die in the narrow region; the obstacle points of interest in the non-narrow region comprise pixel points at an outer edge of the upper die or the lower die which are prone to colliding with the sheet metal part in the non-narrow region.
10 . The robot-assisted sheet metal bending loading and unloading path planning method based on the SVSDF according to claim 9 , wherein in Step 3-6, average_sdf comprises an average average_sdf 1 of the SVSDF of the obstacle points of interest in the narrow region and an average average_sdf 2 of the SVSDF of the obstacle points of interest in the non-narrow region;
in Step 3-7, pdf_value comprises the probability density value pdf_value_narrow of the narrow region and the probability density value pdf_value_nonnarrow of the non-narrow region, and the specific calculation formulae are:
pdf_value
_narrow
=
1
-
1
2
π
σ
1
2
exp
(
-
(
average_sdf
1
-
μ
1
)
2
2
σ
1
2
)
pdf_value
_nonnarrow
=
1
-
1
2
π
σ
2
2
exp
(
-
(
average_sdf
2
-
μ
2
)
2
2
σ
2
2
)
wherein μ 1 is an expected safe distance of the narrow region, which is a set value;
μ 2 is an expected safe distance of the non-narrow region, which is a set value;
σ 1 is a width adjustment value of a normal distribution curve corresponding to the SVSDF value of the obstacle point of interest in the narrow region, which is a set value; and
σ 2 is a width adjustment value of a normal distribution curve corresponding to the SVSDF value of the obstacle point of interest in the non-narrow region, which is a set value.Cited by (0)
No later patents cite this yet.
References (0)
No backward citations on record.