TWI856570B - Processing path planning simulation device and method - Google Patents

Abstract

The processing path planning simulation device is provided, which includes a memory and a processor. The processor performs following operations: according to an obstacle model, multiple processing point positions, a robot arm model, a processing tool model, relationship between models and production strategy parameter, performing collision test simulation to generate multiple candidate poses of the robot arm model; performing path optimization algorithm on multiple candidate poses to generate a pose sequence; performing ant colony algorithm based on the posture sequence and the obstacle model to generate an optimal processing path, where the optimal processing path includes multiple optimal processing segments and multiple optimal nodes, where the ant colony algorithm includes performing multiple path generating operations between positions of multiple optimal postures to generate multiple optimal processing segments; and simulate an end point of a virtual tool on the robot arm model to perform virtual processing operations on the multiple processing point positions according to the multiple optimal nodes in sequence so as to produce simulation results.

Description

Processing path planning simulation device and method

The present disclosure relates to a processing path planning simulation device and method.

In the electronic assembly industry, robotic arms are widely used to replace traditional manual processes due to their flexibility, such as loading and unloading, screwing, glue spraying, assembly and soldering. Most of these processes rely on robotic arm installation tools to perform processing tasks. Furthermore, traditional processing path planning requires engineers to teach points and program on-site, which will occupy actual machines and cause line downtime costs and time costs. In addition, the results of processing path planning are highly dependent on past experience and knowledge. In typical practices, in order to avoid collisions, many auxiliary points are often arranged in the path to bypass obstacles to ensure that the robot arm does not collide with other objects during the execution of the task. In addition, during the process of the robot arm changing hands, the arm posture may change greatly, so this is usually done away from obstacles (for example, after completing a dispensing process, the robot arm is raised 3 to 50 cm to ensure that there is no collision). In this way, the robot arm can move smoothly during the hand-changing operation and avoid collisions to the greatest extent. However, such path planning is often not the best processing path and there is a lot of room for improvement.

The present disclosure provides a processing path planning simulation device, which includes a memory and a processor. The memory stores a plurality of instructions. The processor is connected to the memory to access the plurality of instructions and execute the following operations: performing collision test simulation according to an obstacle model in a virtual environment, a plurality of processing point positions on the obstacle model, a robot arm model in a virtual environment, a processing tool model on the robot arm model, a relative relationship between model positions, and production strategy parameters to generate a plurality of candidate postures of the robot arm model, wherein the plurality of candidate postures are divided into a plurality of posture groups, wherein the plurality of posture groups respectively correspond to the processing point positions; executing a path optimization algorithm on the plurality of candidate postures to generate a posture sequence, wherein the posture sequence The method comprises a plurality of optimal postures arranged in sequence; executing an ant swarm algorithm based on the posture sequence and the obstacle model to generate an optimal processing path, wherein the optimal processing path comprises a plurality of optimal processing segments and a plurality of optimal nodes, wherein the ant swarm algorithm comprises performing a plurality of path generation operations between the positions of the plurality of optimal postures to generate a plurality of optimal processing segments; and on the obstacle model in the virtual environment, based on the optimal processing path, the end point of the processing tool model on the simulated robot arm model sequentially performs virtual processing operations on the positions of the plurality of processing points in postures corresponding to the plurality of optimal nodes to generate a simulation result.

The present disclosure proposes another processing path planning simulation method, including: performing collision test simulation according to an obstacle model in a virtual environment, multiple processing point positions on the obstacle model, a robot arm model in the virtual environment, a processing tool model on the robot arm model, a relative relationship between model positions, and production strategy parameters to generate multiple candidate postures of the robot arm model, wherein the multiple candidate postures are divided into multiple posture groups, wherein the multiple posture groups correspond to the processing point positions respectively; executing a path optimization algorithm on the multiple candidate postures to generate a posture sequence, wherein the posture sequence includes A plurality of optimal postures are arranged in sequence; an ant swarm algorithm is executed based on the posture sequence and the obstacle model to generate an optimal processing path, wherein the optimal processing path includes a plurality of optimal processing segments and a plurality of optimal nodes, wherein the ant swarm algorithm includes performing a plurality of path generation operations between the positions of the plurality of optimal postures to generate a plurality of optimal processing segments; and on the obstacle model in the virtual environment, based on the optimal processing path, the end point of the processing tool model on the simulated robot arm model sequentially performs virtual processing operations on the plurality of processing point positions with postures corresponding to the plurality of optimal nodes to generate a simulation result.

In the past, a lot of downtime and time costs were often incurred. In order to avoid obstacles, many auxiliary points had to be arranged to bypass them, and the robot arm would be far away from the processing point during the hand-changing process, which would make the planned path not optimal (that is, the processing path was not the shortest, and the overall time was not the shortest). These are all problems caused by relying on a lot of human judgment.

In view of this, the present disclosure proposes a processing path planning simulation device and method, which allows the user to first input an obstacle model, a robot arm model, a processing tool model, a relative relationship between model positions, and production strategy parameters to simulate a real field, and then perform a collision test simulation and an ant swarm algorithm based on the obstacle model, the robot arm model, the processing tool model, the relative relationship between model positions, and production strategy parameters to generate an optimal processing path.

This approach can solve the problem that the result of manually planned paths is often not optimal (for example, the path from one processing point to another is too far and not efficient enough). Furthermore, the processing path planning simulation device and method disclosed in the present disclosure can simulate the processing path in advance, which can greatly shorten the downtime cost and time cost caused by the time of manual teaching points and programming, which will improve the result of manually planned paths, shorten the processing path and shorten the processing time. In addition, the path generation operation and collision detection mechanism are combined in the recursive stage of the ant colony algorithm, which will greatly reduce the time required for calculation.

Please refer to FIG. 1, which is a block diagram of a processing path planning simulation device 100 according to some embodiments of the present disclosure. As shown in FIG. 1, the processing path planning simulation device 100 includes a memory 110 and a processor 120. The memory 110 and the processor 120 are connected to each other.

In some embodiments, the processing path planning simulation device 100 can be implemented by an electronic device such as a smart phone, a tablet computer, a laptop computer, a desktop computer, a relay device, or a server.

In this embodiment, the memory 110 stores instructions executed by the processor 120, which are used to execute the detailed steps described in the following paragraphs. In some embodiments, the memory 110 can be implemented by a memory unit, a flash memory, a read-only memory, a hard disk, or any equivalent storage component. In some embodiments, the processor 120 can be implemented by a processing unit, a central processing unit, or a computing unit.

Referring to FIG. 2, FIG. 2 is a flow chart of a processing path planning simulation method according to some embodiments of the present disclosure. The components in the processing path planning simulation device 100 of FIG. 1 are used to execute steps S210 to S240 in the processing path planning simulation method. As shown in FIG. 2, first, in step S210, a collision test simulation is performed based on an obstacle model in a virtual environment, a plurality of processing point positions on the obstacle model, a robot model in a virtual environment, a processing tool model on the robot model, a relative relationship between model positions, and production strategy parameters to generate a plurality of candidate postures of the robot model, wherein the plurality of candidate postures are divided into a plurality of posture groups, wherein the plurality of posture groups respectively correspond to a plurality of processing point positions.

In some embodiments, the processing path planning simulation device 100 may further include an input interface (not shown), and an obstacle model in a virtual environment, multiple processing point positions on the obstacle model, a robot arm model in the virtual environment, a processing tool model on the robot arm model, a relative relationship between model positions, and production strategy parameters may be input through the input interface.

In some embodiments, the obstacle model can be a three-dimensional stereoscopic model in a three-dimensional virtual environment, and can be a three-dimensional stereoscopic model of a physical object to be processed (e.g., a printed circuit board (PCB)). For example, a 3D CAD file of a three-dimensional stereoscopic model of a printed circuit board can be input through a user interface, and a three-dimensional virtual environment in which the three-dimensional stereoscopic model exists can be simulated. In some embodiments, a plurality of processing point positions can be pre-set on the obstacle model. For example, a specific line of the virtualized printed circuit board can be set as a processing point position.

In some embodiments, the robot arm model can be another three-dimensional stereoscopic model in a three-dimensional virtual environment, and can be a three-dimensional stereoscopic model of a real robot arm virtualized. For example, a 3D CAD file of a three-dimensional stereoscopic model of a robot arm can be input through a user interface, and a three-dimensional virtual environment in which the three-dimensional stereoscopic model exists can be simulated. In some embodiments, the processing tool model on the robot arm model can be a virtual processing tool such as a virtual glue spraying tool or a screw locking tool set on the robot arm model in the three-dimensional virtual environment. For example, a 3D CAD file of a three-dimensional solid model of a glue spraying tool or a screw locking tool can be input through the user interface, and a three-dimensional virtual environment containing the three-dimensional solid model can be simulated.

In some embodiments, the model position relative relationship may include an assembly relative relationship between a robot arm model and a processing tool model and a posture relative relationship between an obstacle model and the robot arm model in a three-dimensional virtual environment.

In other words, the obstacle model, the robot arm model, and the processing tool model are all real objects (eg, a panel to be processed and a robot arm with a glue spraying tool) in the simulated real scene (eg, a factory).

Referring to FIG. 3 , FIG. 3 is a schematic diagram of an obstacle model OM in a virtual environment SE according to some embodiments of the present disclosure. As shown in FIG. 3 , the virtual environment SE is a virtual three-dimensional space (i.e., having x, y, and z directions), wherein the virtual environment SE can be divided into a plurality of grids, wherein these grids are adjusted by the object grid resolution, and each grid can indicate a coordinate, such a grid can be beneficial to the coordinate identification of the subsequent collision test simulation. In addition, there is an obstacle model OM in the virtual environment SE. The obstacle model OM is a three-dimensional solid model (for example, the grid of the obstacle model OM will indicate multiple corresponding coordinates). There are multiple virtual processing points PP1~PP4 on the obstacle model OM (for example, the grid of the virtual processing points PP1~PP4 will also indicate multiple corresponding coordinates). The virtual processing points PP1~PP4 are used to simulate the points on the obstacle model OM to be simulated.

Referring to FIG. 4, FIG. 4 is a schematic diagram of a robot arm model MAM in a virtual environment SE in some embodiments of the present disclosure. As shown in FIG. 4, a robot arm model MAM may also exist in the virtual environment SE of FIG. 3 (for example, the grid of the robot arm model MAM indicates multiple corresponding coordinates), and the processing tool model ST may be set at the end of the robot arm model MAM (for example, the grid of the processing tool model ST indicates multiple corresponding coordinates).

In some embodiments, the production strategy parameters may include a collision safety distance, a lateral offset distance, a longitudinal offset distance, a vertical distance difference, a tilt difference, and an azimuth difference. In some embodiments, for each of the plurality of processing point positions, a plurality of collision test positions of the end point of the processing tool model ST on the robot arm model MAM may be generated according to the lateral offset distance, the longitudinal offset distance, the vertical distance difference, the tilt difference, and the azimuth difference. Then, for each of the multiple processing point positions, multiple collision test postures of the robot arm model MAM can be generated according to the multiple collision test positions, and multiple non-collision postures can be selected from the multiple collision test postures according to the collision safety distance, the multiple collision test postures and the obstacle model, wherein the robot arm model MAM indicates the six-dimensional (6 degree of freedom, 6DOF) coordinates (i.e., 3D movement space and 3D rotation space) of multiple reference points on the robot arm model, wherein the multiple reference points include the end points of the processing tool model ST on the robot arm model MAM. Then, the multiple non-collision postures corresponding to each of the multiple processing point positions can be used as multiple candidate postures.

It is worth noting that the present disclosure provides users with input of different process strategies as production strategy parameters, and the paths planned by incorporating these strategy information can be safe and feasible. In some embodiments, the production strategy parameters further include but are not limited to tracking or static processing options, front or back processing options, vertical lifting parameters, and center radiation angle parameters.

In some embodiments, these collision test postures indicate the six-dimensional coordinates of the end points of the processing tool model ST. In some embodiments, the joints of the robot arm model MAM, the mechanical parts connected by the joints, the processing tool model ST, and the end points of the processing tool model ST can be used as reference points, and these reference points all have their own six-dimensional coordinates. In some embodiments, the processing tool model ST can be a virtualized spray tool or a virtualized screw tool, etc. For example, when the processing tool model ST is a virtualized screw tool, for each of the multiple processing point positions, the collision test positions of the end points of the processing tool model ST on the robot arm model MAM can be directly generated at the multiple processing point positions. At this time, since the screw threading tool will not process in a tilted posture, the tilt angle difference of the processing tool model ST can be set to 0, and an azimuth angle difference can be set. In addition, due to the processing characteristics of the screw threading tool, the screw threading tool needs to be in direct contact with the processing point position (i.e., the screw hole), so the horizontal offset distance, longitudinal offset distance, and vertical distance difference of the processing tool model ST can be set to 0.

Referring to FIG. 5A , FIG. 5A is a schematic diagram showing collision test simulation in some embodiments of the present disclosure. As shown in FIG. 5A , first, for the processing point position PP1, there is a horizontal offset distance h and a vertical offset distance v between the end point TP of the processing tool model ST and the processing point position PP1, and the pointing direction of the processing tool model ST is the -Z direction, and this initial position is used as a collision test position. In this way, the posture of the robot arm model MAM can be adjusted according to the collision test position to generate a collision test posture. When the collision test posture can be generated, the simulated processing tool model ST performs virtual processing and determines whether the processing of the processing point position PP1 can be completed. When the processing can be completed, it can be determined whether the processing tool model ST of the robot arm model MAM will collide with the obstacle model OM in the collision test posture. When there is no collision with the obstacle model OM, this collision test posture can be used as a non-collision posture.

In some embodiments, the method for determining whether a collision occurs may include: determining whether the end point TP of the processing tool model ST reaches a collision test posture by inverse kinematics, wherein when the end point TP of the processing tool model ST reaches a collision test posture, it is possible to determine whether multiple links between multiple reference points in the robot arm model MAM and the processing tool model ST collide with the obstacle model OM under the collision test posture. It is worth noting that this approach can ignore the geometric space of the robot arm model MAM itself, and because this detection mechanism has low computational cost, it can be widely used in the iteration of path planning search (i.e., the iteration of the subsequent ant swarm algorithm). In some embodiments, when the end point TP of the processing tool model ST reaches a collision test posture, the body structure of the robot arm model MAM, multiple links between multiple reference points in the robot arm model MAM, and whether these coordinates of the processing tool model ST intersect with the coordinates of the obstacle model OM can be judged under the collision test posture. It is worth noting that this approach can ensure the reliability of the detection results but the amount of calculation is large, and only in the key steps will the accurate detection mechanism be performed (for example, after the subsequent ant swarm algorithm is completed, the accurate detection mechanism can be performed for the entire path found), so that the goals of high efficiency and reliable planning results can be achieved at the same time.

In some embodiments, it is possible to calculate whether the distance between multiple links between multiple reference points in the robot model MAM and one of the processing tool models ST and the robot model MAM and the obstacle model OM is less than or equal to the collision safety distance in the collision test posture. Then, when the distance is less than or equal to the collision safety distance, it can be determined that a collision has occurred.

Next, as shown in Figure 5B, Figure 5B is a partial enlarged schematic diagram of the initial position in Figure 5A. The pointing direction of the processing tool model ST can be adjusted to have an angle difference φ with the vertical extension line CVL (that is, the center extension line L of the processing tool model ST and the vertical extension line CVL passing through the center point of the circle C (parallel to the +Z direction and below the circle C) will have an intersection, and there will be an angle between the center extension line L of the processing tool model ST and the vertical extension line CVL passing through the center point of the circle C), and the end point TP of the processing tool model ST is set on the circumference of the circle C with a radius of h, and then the same method as above is used to determine whether another non-collision posture can be generated, in which the Z axis can be perpendicular to the circular plane of the circle C (that is, perpendicular to the radius h of the circle C).

Next, the end point TP of the processing tool model ST can be rotated along the circle C by an angle difference θ, and the same method as above is used to determine whether another non-collision posture can be generated. Similarly, the next collision test will be rotated by another rotation angle difference θ until the total rotation angle is equal to 360 degrees, and the angle between the pointing direction of the processing tool model ST and the vertical extension line CVL will be increased by the angle difference φ, and the same process will be continued until the pointing direction of the processing tool model ST is adjusted to a vertical angle equal to a preset angle (which can be set by the user in advance) with the vertical extension line CVL. When the total rotation angle is equal to 360 degrees, the vertical angle is equal to the preset angle, and any non-collision posture is generated, the collision test can be terminated.

Next, when the total rotation angle is equal to 360 degrees, the vertical angle is equal to the preset angle, and no non-collision posture is generated, the end point TP of the processing tool model ST can be moved vertically upward by a vertical distance difference d, and the pointing direction of the processing tool model ST is set to the -Z direction, and the collision test is continued in the same manner as above to generate other non-collision postures until the pointing direction of the processing tool model ST is adjusted again to a vertical angle between the vertical extension line CVL equal to the above preset angle. When the total rotation angle is equal to 360 degrees, the vertical angle is equal to the preset angle, and any non-collision posture is generated, the collision test can also be terminated. By analogy, when the total rotation angle is equal to 360 degrees, the vertical angle is equal to the preset angle, and no non-collision posture has been generated, any non-collision posture can be generated by continuously moving upward until N vertical distance differences d are moved, where N can be a preset positive integer. Finally, other non-collision postures can be generated in the same way for other processing point positions PP2~PP4 in Figure 3.

Furthermore, as shown in FIG. 2, in step S220, a path optimization algorithm is executed on multiple candidate postures to generate a posture sequence, wherein the posture sequence includes multiple optimal postures arranged in sequence. In some embodiments, the path optimization algorithm can be a mathematical planning algorithm, a heuristic algorithm, or a meta-heuristic algorithm. In some embodiments, the best postures of each group can be selected from multiple posture groups and the arrangement order of multiple optimal postures can be generated by a mathematical planning algorithm, a heuristic algorithm, or a meta-heuristic algorithm. Then, a posture sequence can be generated according to the multiple optimal postures and the arrangement order.

The following is a further explanation of the posture groups of the present disclosure with practical examples. Referring to FIG. 6, FIG. 6 is a schematic diagram of multiple posture groups GP1 to GP4 shown in some embodiments of the present disclosure. As shown in FIG. 6, posture group GP1 includes multiple candidate postures at positions COD1 to COD3, posture group GP2 includes multiple candidate postures at positions COD4 to COD5, posture group GP3 includes multiple candidate postures at positions COD6 to COD8, and posture group GP4 includes multiple candidate postures at positions COD9 to COD11, wherein posture groups GP1 to GP4 correspond to processing point positions PP1 to PP4 of FIG. 3, respectively. Specifically, posture group GP1 is composed of candidate postures of positions COD1 to COD3 around processing point position PP1, posture group GP2 is composed of candidate postures of positions COD4 to COD5 around processing point position PP2, posture group GP3 is composed of candidate postures of positions COD6 to COD8 around processing point position PP3, and posture group GP4 is composed of candidate postures of positions COD9 to COD11 around processing point position PP4.

Referring to FIG. 7, FIG. 7 is a schematic diagram of a generated posture sequence illustrated in some embodiments of the present disclosure. As shown in FIG. 7, a candidate posture can be individually selected from the posture groups GP1 to GP4 by a mathematical planning algorithm, a heuristic algorithm, or a meta-heuristic algorithm, and the combination of these candidate postures will have the shortest path sum. Specifically, when a mathematical planning algorithm, a heuristic algorithm, or a meta-heuristic algorithm is performed on the posture groups GP1 to GP4, it can be calculated that a path generated by position COD3, position COD5, position COD8, and position COD9 will have the shortest path sum. Therefore, the candidate postures of position COD3, position COD5, position COD8, and position COD9 can be used as multiple optimal postures arranged in sequence, and the multiple optimal postures arranged in sequence can be used as a posture sequence. In other words, the robot arm model MAM is to use the candidate postures of position COD3, position COD5, position COD8, and position COD9 in sequence to simulate the processing of the processing point positions PP1~PP4 in Figure 3.

Furthermore, in step S230, the ant colony optimization (ACO) algorithm is executed based on the posture sequence and the obstacle model OM to generate an optimal processing path, wherein the optimal processing path includes multiple optimal processing segments and multiple optimal nodes, wherein the ant colony algorithm includes performing multiple path generation operations between the positions of multiple optimal postures to generate multiple optimal processing segments.

In some embodiments, for the positions of two adjacent ones arranged in multiple optimal postures, multiple path generation operations are performed to generate multiple candidate processing segments and multiple candidate nodes, and the ant swarm algorithm is executed according to the ant quantity parameter and the number of repetitions to select multiple optimal processing segments and multiple optimal nodes from the multiple candidate processing segments and multiple candidate nodes. In some embodiments, the number of repetitions indicates the total number of times the multiple repetition stages are executed. In some embodiments, the path generation operation may include: generating multiple candidate processing segments and multiple candidate nodes according to the positions of two adjacent ones arranged in multiple optimal postures, the collision safety distance, and multiple coordinates corresponding to the obstacle model OM.

The following is a practical example to further illustrate the generation of the optimal processing section of the present disclosure. Referring to FIG. 8, FIG. 8 is a schematic diagram of multiple optimal posture positions BP1~BP4 shown in some embodiments of the present disclosure. As shown in FIG. 8, the optimal posture positions BP1~BP4 are respectively located near the processing point positions PP1~PP4. For the convenience of explanation, the following first describes the generation of the optimal processing section between the optimal posture positions BP1~BP2.

Referring to FIG. 9A, FIG. 9A is a schematic diagram of the first step in the first selection between two adjacent positions in multiple optimal postures shown in some embodiments of the present disclosure. As shown in FIG. 9A, there are two simulated obstacles OS1 and OS2 near the positions BP1 and BP2 of the optimal posture, and the optimal processing section is to be found between the position BP1 of the optimal posture and the position BP2 of the optimal posture. At this time, one of the recursive stages can be performed.

In this recursive stage, four candidate processing segments PS11 to PS14 (assuming that four are generated each time) and three corresponding candidate nodes ND11 to ND13 can be generated by path generation operation. In addition, the end point TP of the processing tool model ST on the robot arm model MAM can be identified by the same method as the above-mentioned collision test, which will not collide with the simulated obstacles OS1 and OS2 on the candidate processing segments PS11 to PS13 and the candidate nodes ND11 to ND13 (that is, the candidate processing segment PS14 will collide). Assuming that the ant number parameter is set to 3, the number of repetitions is set to 4, and the initial pheromone concentration of the candidate processing sections PS11-PS13 is set to τ ₀ (a preset constant), one of the candidate processing sections PS11-PS13 can be selected in the first selection of the first repetition (i.e., the first ant) based on a selection probability, where the selection probability is shown in the following formula (1).

為第k隻螞蟻從候選節點i到候選節點j的機 率，τ _ij 是位於候選節點i、j之間的路徑上的費洛蒙濃度(此時所有候選加工段的費洛蒙濃度都是τ₀)，η _ij 為期望值，其為候選節點i、j之間的最短路徑長度之倒數(例如，從最佳姿態的位置BP1到候選節點ND11之間的最短路徑長度的倒數)，z為從候選節點i可能到達的所有候選節點(即，從最佳姿態的位置BP1可能到達候選節點ND11~ND13)，α是費洛蒙指數項，β是距離倒數指數項，α,β用以決定費洛蒙與距離相對重要性之參數。此時越短的路徑會有越高的選擇機率(如第9A圖中的候選加工段PS12)。 in

is the probability of the kth ant going from candidate node i to candidate node j , τ _ij is the pheromone concentration on the path between candidate nodes i and j (the pheromone concentration of all candidate processing sections is τ ₀ at this time), η _ij is the expected value, which is the reciprocal of the shortest path length between candidate nodes i and j (for example, the reciprocal of the shortest path length from the best posture position BP1 to the candidate node ND11), z is all the candidate nodes that can be reached from candidate node i (that is, from the best posture position BP1, candidate nodes ND11~ND13 can be reached), α is the pheromone index term, β is the reciprocal distance index term, α and β are parameters used to determine the relative importance of pheromone and distance. At this time, the shorter the path, the higher the chance of selection (such as the candidate processing section PS12 in Figure 9A).

Referring to FIG. 9B, FIG. 9B is a schematic diagram of the second step in the first selection between two positions arranged adjacent to each other in a plurality of optimal postures according to some embodiments of the present disclosure. As shown in FIG. 9B, when the candidate processing segment PS12 is selected, four candidate processing segments PS121-PS124 and corresponding four candidate nodes ND121-ND124 can be generated by path generation operation. In addition, the end point TP of the processing tool model ST on the robot arm model MAM can be identified by the same method as the above-mentioned collision test, which will not collide with the simulated obstacles OS1 and OS2 on the candidate processing segments PS121-PS122 and the candidate nodes ND121-ND122 (that is, the candidate processing segments PS123-PS124 will collide). Then, the selection probability can be calculated based on the same method as above to select one of the candidate processing sections PS121~PS122.

Referring to FIG. 9C , FIG. 9C is a schematic diagram of the third step in the first selection between two positions arranged adjacent to each other in a plurality of optimal postures, illustrated in some embodiments of the present disclosure. As shown in FIG. 9C , when the candidate processing segment PS122 is selected, four candidate processing segments PS1221 to PS1224 and corresponding four candidate nodes ND1221 to ND1224 can be generated by path generation operation. In addition, the end point TP of the processing tool model ST on the robot arm model MAM can be identified by the same method as the above-mentioned collision test, so that it will not collide with the simulated obstacles OS1 and OS2 on the candidate processing segments PS1221 to PS1222 and the candidate nodes ND1221 to ND1223 (that is, the candidate processing segment PS1224 will collide). Then, the selection probability can be calculated based on the same method as above to select one of the candidate processing sections PS1221~PS1223.

Referring to FIG. 9D, FIG. 9D is a schematic diagram of multiple candidate processing segments PS11~PS13, PS121~PS122, PS1221~PS1223 generated by the first selection between two adjacent positions in multiple optimal postures according to some embodiments of the present disclosure. As shown in FIG. 9D, at this time, candidate processing segments PS11~PS13, candidate processing segments PS121~PS122, and candidate processing segments PS1221~PS1223 can be generated in the first selection. When candidate processing segment PS1223 is selected and the position BP2 of the optimal posture is reached, the path of the first selection may include candidate processing segments PS12, PS122, and PS1223.

Referring to FIG. 10A , FIG. 10A is a schematic diagram of one step of the second selection between two positions arranged adjacent to each other in a plurality of optimal postures according to some embodiments of the present disclosure. As shown in FIG. 10A , assuming that in the second selection of the first recursion, the selection machine first selects the candidate processing segments PS12 and PS121 based on the previous calculation, four candidate processing segments PS1211 to PS1214 and corresponding three candidate nodes ND1211 to ND1212 and ND1214 can be generated by path generation operation. In addition, the same method as the above collision test can be used to identify that the end point TP of the processing tool model ST on the robot arm model MAM will not collide with the simulated obstacles OS1 and OS2 on the candidate processing segments PS1211~PS1213 and the candidate nodes ND1211~ND1212 (that is, the candidate processing segment PS1214 will collide). Then, the selection probability can be calculated based on the same method as above to select one of the candidate processing segments PS1211~PS1213.

Referring to FIG. 10B, FIG. 10B is a schematic diagram of multiple candidate processing segments PS11~PS13, PS121~PS122, PS1211~PS1213, PS1221~PS1223 generated by the second selection between two adjacent positions in multiple optimal postures according to some embodiments of the present disclosure. As shown in FIG. 10B, at this time, candidate processing segments PS1211~PS1213 can be generated in the second selection. When candidate processing segment PS1213 is selected and the position BP2 of the optimal posture is reached, the path of the second selection may include candidate processing segments PS12, PS121 and PS1213.

Referring to FIG. 11A , FIG. 11A is a schematic diagram of one step of the third selection between two positions arranged adjacent to each other in a plurality of optimal postures according to some embodiments of the present disclosure. As shown in FIG. 11A , assuming that the selection machine first selects the candidate processing segment PS13 based on the previously calculated selection in the third selection of the first recursion, four candidate processing segments PS131 to PS134 and corresponding three candidate nodes ND131 to ND132 and ND134 can be generated by path generation operation. In addition, the same method as the above collision test can be used to identify that the end point TP of the processing tool model ST on the robot arm model MAM will not collide with the simulated obstacles OS1 and OS2 on the candidate processing segments PS133~PS134 and the candidate node ND134 (that is, the candidate processing segments PS131~PS132 will collide). Then, the selection probability can be calculated based on the same method as above to select one of the candidate processing segments PS133~PS134.

Referring to FIG. 11B, FIG. 11B is a schematic diagram of multiple candidate processing segments PS11~PS13, PS121~PS122, PS1211~PS1213, PS1221~PS1223, PS133~PS134 generated by the third selection between two positions arranged adjacent to each other in multiple optimal postures according to some embodiments of the present disclosure. As shown in FIG. 11B, at this time, candidate processing segments PS133~PS134 can be generated in the third selection. When candidate processing segment PS133 is selected and reaches the position BP2 of the optimal posture, the path of the third selection may include candidate processing segments PS13 and PS133. Then, after the third selection (i.e., the first recursion ends), the pheromone concentration of all the above-mentioned candidate processing segments can be set as shown in the following formula (2).

為第k隻螞蟻(即，第k次選擇路徑)在候選節點i、j之間的路徑上所釋放的費洛蒙數量，且設為第k次選擇路徑總長度的倒數與常數Q的乘積，其中ρ以及Q皆為預先設定好的常數。此外，若第k次沒有選擇經過候選節點i、j之間，則為0，不會在其上釋放費洛蒙。換言之，每次遞迴結束都可利用公式(2)更新所有候選加工段的費洛蒙濃度。接著，之後的2~4次遞迴也是以相同的方式產生並選擇路徑以及更新費洛蒙濃度。 Where τ _ij is the pheromone concentration on the path between candidate nodes i and j (the pheromone concentration of all candidate processing sections is τ ₀ at this time), t is the t-th recurrence, ρ is the evaporation coefficient of pheromone,

is the amount of pheromone released by the kth ant (i.e., the kth selected path) on the path between candidate nodes i and j , and is set to the product of the reciprocal of the total length of the kth selected path and the constant Q, where ρ and Q are both pre-set constants. In addition, if the kth ant does not choose to pass through the candidate nodes i and j , it is 0 and no pheromone will be released on it. In other words, at the end of each recursion, the pheromone concentration of all candidate processing sections can be updated using formula (2). Then, the next 2 to 4 recursions are also generated and the path is selected and the pheromone concentration is updated in the same way.

In some embodiments, the processing path planning simulation device 100 may further include a user interface (not shown), and the simulation results may be displayed through the user interface. In some embodiments, the user interface may be implemented by various display circuits such as a display or a projector.

Refer to FIG. 12, which is a schematic diagram of simulation results shown in some embodiments of the present disclosure. As shown in FIG. 12, by the method of FIG. 9A to FIG. 11B, multiple optimal processing segments BPS1 to BPS5 can be generated between the positions BP1 to BP4 of the optimal posture. In addition, in addition to the positions BP1 to BP4 of the optimal posture, nodes ND1 to ND2 are also generated, and two other operating postures of the robot arm model MAM are generated at the nodes ND1 to ND2. Thus, the positions BP1 to BP4 of the optimal posture and the nodes ND1 to ND2 can be used as multiple optimal nodes, and multiple optimal processing segments BPS1 to BPS5 and multiple optimal nodes can form an optimal processing path, which not only has the shortest path length but also can avoid collisions with obstacles.

In some embodiments, the optimal processing path is adjusted according to the simulation result, wherein the adjusted optimal processing path includes multiple adjusted optimal processing segments and multiple adjusted optimal nodes. Then, on the obstacle model OM in the virtual environment SE, based on the adjusted optimal processing path, the end point TP of the processing tool model ST on the simulated robot arm model MAM sequentially performs virtual processing operations on multiple processing point positions PP1-PP4 in postures corresponding to the multiple adjusted optimal nodes to generate adjusted simulation results.

In some embodiments, a path pruning algorithm can be used to delete the best node between two best processing segments in the best processing path to generate an adjusted best processing segment. For example, three adjacent best nodes between multiple best processing segments can be randomly selected, and the front and rear best nodes can be connected to generate a test processing segment, and the above-mentioned collision test can be performed on this test processing segment. If no collision occurs, the middle best node among the three adjacent best nodes can be deleted, and this test processing segment can be used as the adjusted best processing segment to generate an adjusted best processing path. In some embodiments, the best processing path can be smoothed by a smoothing algorithm such as a Bézier curve algorithm or a B-spline algorithm.

In some embodiments, a path file may be generated based on the optimal processing path, wherein the path file indicates the posture of the robot arm model MAM at multiple optimal nodes. The following table (1) shows an example of a path file.

As shown in Table (1) above, the path file may include the six-dimensional coordinates of multiple optimal nodes. Thus, the user can view the path file through the user interface.

In summary, the processing path planning simulation device and method disclosed in the present invention first input various simulation parameters and models to determine the best posture for simulating the processing of the processing point position, and then determine the simulation processing order of these best postures, and then combine the ant colony algorithm and path generation operation to generate the best processing path between the best postures. In this way, as long as the required parameters and models are input, the best processing path on the physical object to be processed can be automatically simulated. Therefore, the line stop cost and time cost caused by on-site teaching points and programming can be solved, and the problem that the planned path is not optimal can be further solved.

Although the present disclosure has been disclosed as above by way of embodiments, it is not intended to limit the present disclosure. Any person with ordinary knowledge in the relevant technical field may make some changes and modifications within the spirit and scope of the present disclosure. Therefore, the protection scope of the present disclosure shall be subject to the scope defined by the attached patent application.

100: Processing path planning simulation device

110: Memory

120: Processor

S210~S240: Steps

SE: Virtual Environment

PP1~PP4: Processing point location/virtual processing point

OM: Obstacle Model

MAM: Robotic Arm Model

θ: rotation angle difference

φ: Tilt difference

CVL:Vertical extension line

ST: Processing tool model

h: lateral offset distance

v: longitudinal offset distance

d: vertical distance difference

TP: terminal point

L: Center extension line

C: Round

GP1~GP4: Posture group

COD1~COD11: Location

BP1~BP4: The best posture position

OS1~OS2: Simulated obstacles

PS11~PS14,PS121~PS124,PS1211~PS1214,PS1221~PS1224,PS131~PS134: Candidate processing sections

ND11~ND13,ND121~ND124,ND1211,ND1212,ND1214,ND1221~ND1224,ND131,ND132,ND134: candidate nodes

ND1~ND2: Node

BPS1~BPS5: Optimal processing section

FIG. 1 is a block diagram of a processing path planning simulation device illustrated in some embodiments of the present disclosure. FIG. 2 is a flow chart of a processing path planning simulation method illustrated in some embodiments of the present disclosure. FIG. 3 is a schematic diagram of an obstacle model in a virtual environment illustrated in some embodiments of the present disclosure. FIG. 4 is a schematic diagram of a robot arm model in a virtual environment illustrated in some embodiments of the present disclosure. FIG. 5A is a schematic diagram of a collision test simulation illustrated in some embodiments of the present disclosure. FIG. 5B is a schematic diagram of a partial enlarged view of the initial position in FIG. 5A. FIG. 6 is a schematic diagram of multiple posture groups illustrated in some embodiments of the present disclosure. FIG. 7 is a schematic diagram of a generated posture sequence illustrated in some embodiments of the present disclosure. FIG. 8 is a schematic diagram of the positions of multiple optimal postures shown in some embodiments of the present disclosure. FIG. 9A is a schematic diagram of the first step in the first selection between positions arranged adjacent to each other in multiple optimal postures shown in some embodiments of the present disclosure. FIG. 9B is a schematic diagram of the second step in the first selection between positions arranged adjacent to each other in multiple optimal postures shown in some embodiments of the present disclosure. FIG. 9C is a schematic diagram of the third step in the first selection between positions arranged adjacent to each other in multiple optimal postures shown in some embodiments of the present disclosure. FIG. 9D is a schematic diagram of multiple candidate processing sections generated by the first selection between positions arranged adjacent to each other in multiple optimal postures shown in some embodiments of the present disclosure. FIG. 10A is a schematic diagram of one step of the second selection between two positions arranged adjacent to each other in a plurality of optimal postures, as shown in some embodiments of the present disclosure. FIG. 10B is a schematic diagram of multiple candidate processing segments generated by the second selection between two positions arranged adjacent to each other in a plurality of optimal postures, as shown in some embodiments of the present disclosure. FIG. 11A is a schematic diagram of one step of the third selection between two positions arranged adjacent to each other in a plurality of optimal postures, as shown in some embodiments of the present disclosure. FIG. 11B is a schematic diagram of multiple candidate processing segments generated by the third selection between two positions arranged adjacent to each other in a plurality of optimal postures, as shown in some embodiments of the present disclosure. FIG. 12 is a schematic diagram of simulation results, as shown in some embodiments of the present disclosure.

S210~S240: Steps

Claims (10)

A processing path planning simulation device includes: a memory storing a plurality of instructions; a processor connected to the memory for accessing the instructions and performing the following operations: performing a collision test simulation based on an obstacle model in a virtual environment, a plurality of processing point positions on the obstacle model, a robot arm model in the virtual environment, a processing tool model on the robot arm model, a relative relationship between model positions and a production strategy parameter to generate a plurality of processing point positions of the robot arm model; candidate postures, wherein the collision test simulation is to simulate the multiple non-collision postures of the robot arm model in which the processing tool model will not collide with the obstacle model when the processing tool model processes each of the processing point positions, and the non-collision postures corresponding to each of the processing point positions are used as the candidate postures, wherein the candidate postures are divided into a plurality of posture groups; a path optimization algorithm is executed on the candidate postures to generate a posture sequence, wherein the posture sequence includes A plurality of optimal postures are arranged in sequence, wherein the posture groups correspond to a plurality of optimal positions, respectively, the optimal positions correspond to the processing point positions and are respectively around the processing point positions, a path generated by the optimal positions will have the shortest path sum, and the candidate postures corresponding to the optimal positions are used as the optimal postures arranged in sequence; an ant swarm algorithm is executed based on the posture sequence and the obstacle model to generate an optimal processing path, wherein the optimal processing The path includes multiple optimal processing segments and multiple optimal nodes, wherein the ant colony algorithm includes generating the optimal processing segments by performing multiple path generation operations between the optimal positions of the optimal postures; and on the obstacle model in the virtual environment, based on the optimal processing path, simulating an end point of a processing tool model on the robot arm model to perform virtual processing operations on the processing point positions in sequence with postures corresponding to the optimal nodes to generate a simulation result. A processing path planning simulation device as described in claim 1, wherein the production strategy parameters include a collision safety distance, a lateral offset distance, a longitudinal offset distance, a vertical distance difference, an inclination difference, and an azimuth difference, wherein the processor is further used to perform the following operations: for each of the processing point positions, generate multiple collision test positions of the end point of the processing tool model on the robot arm model according to the lateral offset distance, the longitudinal offset distance, the vertical distance difference, the inclination difference, and the azimuth difference; for each of the processing point positions, generate multiple collision test positions of the end point of the processing tool model on the robot arm model according to the lateral offset distance, the longitudinal offset distance, the vertical distance difference, the inclination difference, and the azimuth difference; For each of the processing point positions, a plurality of collision test postures of the robot model are generated according to the collision test positions, and the non-collision postures are selected from the collision test postures according to the collision safety distance, the collision test postures and the obstacle model, wherein the robot model indicates the six-dimensional coordinates of a plurality of reference points on the robot model, wherein the reference points include an end point of a processing tool model on the robot model; and the non-collision postures corresponding to each of the processing point positions are used as the candidate postures. A processing path planning simulation device as described in claim 1, wherein the path optimization algorithm is a mathematical planning algorithm, a heuristic algorithm or a meta-heuristic algorithm, and wherein the processor is further used to perform the following operations: selecting the respective best postures from the posture groups and generating a sequence of the best postures by the mathematical planning algorithm, the heuristic algorithm or the meta-heuristic algorithm; and generating the posture sequence according to the best postures and the sequence. The processing path planning simulation device as described in claim 2, wherein the processor is further used to perform the following operations: performing the path generation operations according to the optimal positions of the two adjacent ones arranged in the optimal postures, the collision safety distance, and the multiple coordinates corresponding to the obstacle model to generate multiple candidate processing segments and multiple candidate nodes, and executing the ant swarm algorithm according to the ant quantity parameter and the number of iterations to select multiple optimal processing segments and multiple optimal nodes from the candidate processing segments and the candidate nodes. The processing path planning simulation device as described in claim 1, wherein the processor is further used to perform the following operations: adjusting the optimal processing path according to the simulation result, wherein the adjusted optimal processing path includes a plurality of adjusted optimal processing segments and a plurality of adjusted optimal nodes; and on the obstacle model in the virtual environment, based on the adjusted optimal processing path, simulating the end point of the processing tool model on the robot arm model to perform the virtual processing operation on the processing point positions in sequence with the posture corresponding to the adjusted optimal nodes to generate the adjusted simulation result. A processing path planning simulation method includes: performing a collision test simulation based on an obstacle model in a virtual environment, multiple processing point positions on the obstacle model, a robot arm model in the virtual environment, a processing tool model on the robot arm model, a relative relationship between model positions, and a production strategy parameter to generate multiple candidate postures of the robot arm model, wherein the collision test simulation is to simulate the processing tool model to the processing points. When processing each of the point positions, the processing tool model will not collide with the obstacle model, and the non-collision postures corresponding to each of the processing point positions are used as the candidate postures, wherein the candidate postures are divided into a plurality of posture groups; a path optimization algorithm is executed on the candidate postures to generate a posture sequence, wherein the posture sequence includes a plurality of optimal postures arranged in sequence, wherein the postures The groups correspond to a plurality of optimal positions, respectively, and the optimal positions correspond to the processing point positions and are respectively around the processing point positions. A path generated by the optimal positions will have the shortest path sum. The candidate postures corresponding to the optimal positions are used as the optimal postures arranged in sequence. An ant swarm algorithm is executed based on the posture sequence and the obstacle model to generate an optimal processing path, wherein the optimal processing path includes multiple An optimal processing segment and multiple optimal nodes, wherein the ant colony algorithm includes generating the optimal processing segments by performing multiple path generation operations between the optimal positions of the optimal postures; and on the obstacle model in the virtual environment, based on the optimal processing path, simulating an end point of a processing tool model on the robot arm model to perform virtual processing operations on the processing point positions in sequence with the optimal nodes to generate a simulation result. A processing path planning simulation method as described in claim 6, wherein the production strategy parameters include a collision safety distance, a lateral offset distance, a longitudinal offset distance, a vertical distance difference, a tilt difference, and an azimuth difference, wherein the step of performing a collision test simulation to generate the candidate postures of the robot model according to the obstacle model in the virtual environment, the processing point positions on the obstacle model, the robot arm model in the virtual environment, and the production strategy parameters includes: for each of the processing point positions, according to the lateral offset distance, the longitudinal offset distance, the vertical distance difference, the tilt difference, the azimuth difference, The invention relates to a method for generating a plurality of collision test positions of the end point of the processing tool model on the robot arm model by using a difference; generating a plurality of collision test postures of the robot arm model according to the collision test positions for each of the processing point positions, and selecting the non-collision postures from the collision test postures according to the collision safety distance, the collision test postures and the obstacle model, wherein the robot arm model indicates the six-dimensional coordinates of a plurality of reference points on the robot arm model, wherein the reference points include an end point of a processing tool model on the robot arm model; and taking the non-collision postures corresponding to the processing point positions as the candidate postures. A processing path planning simulation method as described in claim 6, wherein the path optimization algorithm is a mathematical planning algorithm, a heuristic algorithm or a meta-heuristic algorithm, wherein the step of executing the path optimization algorithm on the candidate postures to generate the posture sequence includes: selecting the respective best postures from the posture groups and generating a permutation sequence of the best postures by the mathematical planning algorithm, the heuristic algorithm or the meta-heuristic algorithm; and generating the posture sequence according to the best postures and the permutation sequence. The processing path planning simulation method as described in claim 7, wherein the step of executing the ant swarm algorithm based on the posture sequence and the obstacle model to generate the optimal processing path includes: performing the path generation operations according to the optimal positions of the two adjacent ones arranged in the optimal postures, the collision safety distance, and the multiple coordinates corresponding to the obstacle model to generate multiple candidate processing segments and multiple candidate nodes, and executing the ant swarm algorithm according to the ant quantity parameter and the number of iterations to select multiple optimal processing segments and multiple optimal nodes from the candidate processing segments and the candidate nodes. The processing path planning simulation method as described in claim 6 further includes: adjusting the optimal processing path according to the simulation result, wherein the adjusted optimal processing path includes a plurality of adjusted optimal processing segments and a plurality of adjusted optimal nodes; and on the obstacle model in the virtual environment, based on the adjusted optimal processing path, simulating the end point of the processing tool model on the robot arm model to sequentially perform the virtual processing operation on the processing point positions with the posture corresponding to the adjusted optimal nodes to generate the adjusted simulation result.

Images