DE102017200015A1 - Determine at least one subsequent record for a real-time application - Google Patents

Abstract

The invention relates to a method for determining at least one subsequent data record (30) for a real-time application, comprising determining at least two preceding data records (10, 20), the at least two data records (10, 20) each having a grid structure with a multiplicity of nodes (10). 12, 22), comparing the respective nodes (12) of the one grid structure of the at least two grid structures with the respective nodes (22) of the other grid structure of the at least two grid structures, if two nodes (12, 22) match, the identical node (12) 12, 22, 32) is taken over into the subsequent data set if two nodes (12, 22) do not match, the node of the one grid structure of the at least two grid structures is assigned to a corresponding node (32) and the assigned node (32) in the subsequent record (30) is taken. Furthermore, the invention is directed to a corresponding computer program product.

Description

Technical area

The present invention relates to a method for determining at least one subsequent data record for a real-time application. Furthermore, the invention is directed to a corresponding computer program product.

State of the art

Virtual Reality, or Virtual Reality (VR) called, is becoming increasingly important. Virtual reality is the representation of a virtual world, which is computer-generated. For example, the virtual world may be based on the real world.

Furthermore, virtual reality is closely related to VR glasses. By using a VR glasses, a user is advantageously no longer dependent on a monitor, but has a 3D image in his field of vision, which can be conveniently directed in all directions to view the virtual world.

Usually, animations or illustrations, each with 30 to 50 frames per second, have to be displayed in order to be perceived as a movement for the human eye. It is necessary that the individual images (also referred to as "frame") run as steadily as possible and that no large jumps occur between the individual images.

Such jumps easily cause irritation, dizziness, headache and vomiting of the user of VR glasses. These symptoms are also summarized under the term "Virtual Reality Sickness".

3 illustrates an exemplary method for processing a sequence of computational results that are iteratively / successively into an animation with individual images and their disadvantageous jumps according to the prior art. The individual method steps are shown sequentially from top to bottom and will be described in the following Detail explains:

First, in the top row, the computational results of the iterations i-1, ..., i + 2 are entered, which are further processed into individual images in the last row. The iterations i-1, ..., i + 2 are to be understood as calculating steps. The duration of an iteration may vary and may take several seconds, for example. An example of an iterative method is a shape optimization process in which, based on predefined forces, an optimized structure is calculated which withstands the forces and additionally has the lowest possible weight. The structure can be displayed to the user or simulation engineer during the calculation. Another example would be any FEM (Finite Element Method) calculation, e.g. a fluid simulation or a thermal simulation.

In the second row, the calculation results of individual iterations i-1,..., I + 2 are graphically illustrated, for example, as a rectangle, a circle and a triangle. The calculation results can be understood as data to be illustrated in the animation. Only the calculation result of the previous iteration is known, but not the result of the calculation of the subsequent or future iteration. In the shape optimization process, for example, after each iteration, an improved (mesh) structure is output, which is an input of the next iteration and can also be graphically illustrated.

The individual computation results are further processed into a temporally continuous sequence of computation results. Accordingly, in the third row the further processed calculation results are graphically illustrated. In the prior art, the previous calculation result is simply used as an intermediate result between two successive computation results, namely the preceding - known - computation result and the following - unknown - computation result. The still image resulting from the calculation result is displayed to the user in the animation in the last row.

Due to the still images between the frames, however, disadvantageous jumps occur between the individual images of the animation, as shown in the last lower row. Consequently, the animation is unsteady and it may also come for the user to the above-mentioned "Virtual Reality Sickness".

The present invention therefore has as its object to provide a method for determining at least one subsequent data record for a real-time application, which enables a continuous animation of the resulting individual images.

Summary of the invention

1. a. Ermitteln mindestens zweier vorangegangener Datensätze, wobei
2. b. die mindestens zwei vorangegangenen Datensätze jeweils eine Gitterstruktur mit einer Vielzahl von Knoten aufweisen,
3. c. Vergleichen der jeweiligen Knoten der einen Gitterstruktur der mindestens zwei Gitterstrukturen mit den jeweiligen Knoten der anderen Gitterstruktur der mindestens zwei Gitterstrukturen,
   • - falls zwei Knoten übereinstimmen, der identische Knoten in den nachfolgenden Datensatz übernommen wird,
   • - falls zwei Knoten nicht übereinstimmen, der Knoten der einen Gitterstruktur der mindestens zwei Gitterstrukturen einem entsprechenden Knoten zugewiesen wird und der zugewiesene Knoten in den nachfolgenden Datensatz übernommen wird.

The above object is achieved according to the invention by a method for determining at least one subsequent data record for a real-time application, comprising:

1. a. Determine at least two previous records, where
2. b. the at least two preceding data records each have a grid structure with a multiplicity of nodes,
3. c. Comparing the respective nodes of the one grid structure of the at least two grid structures with the respective nodes of the other grid structure of the at least two grid structures,
   • if two nodes match, the identical node is taken over into the subsequent data record,
   • if two nodes do not match, the node of the one grid structure of the at least two grid structures is assigned to a corresponding node and the assigned node is taken over into the subsequent data record.

Accordingly, a subsequent data record is determined by the method according to the invention. The determination of the subsequent data record is based on two or more data records which precede or precede the subsequent data record to be determined. Accordingly, these records may also be referred to as previous records and are known.

The preceding and following data records each have a grid structure. The lattice structure is also referred to as a "mesh structure" or lattice set and has a plurality of nodes. Accordingly, the lattice structures of the previous data sets can be defined as M _i with N _i nodes and M _i-1 with N _i-1 nodes. N corresponds to the number of nodes at the preceding times i and i-1.

The node is defined as the intersection between two or more gridlines in a grid structure. For example, for nodes, the node has information in the form of Cartesian coordinates about the position of the node (X, Y, Z). Furthermore, the node may have further color-coded properties, such as temperature (T), etc. Accordingly, the node can also be viewed from multidimensional vector (X, Y, Z, T).

The respective nodes of the one grid structure of the at least two grid structures are compared with the respective nodes of the other grid structure of the at least two grid structures to determine the subsequent data record. Accordingly, each node of one lattice structure is compared to each other node of the other lattice structure. In other words, one vector (X ₀ , Y ₀ , Z ₀ , T ₀ ) is compared with another vector (X ₁ , Y ₁ , Z ₁ , T ₁ ). The comparison can also be regarded as an actual / target comparison.

The respective nodes are first compared for agreement. Two nodes can be identical or identical to each other. If two nodes are identical, the identical node is taken over into the subsequent data record. This does not require animation. In other words, those nodes are first recognized that do not change between two computation results. These identical nodes may also be referred to as unchanged nodes.

If two nodes are not identical, an assignment takes place and the assigned node is taken over into the subsequent data record. In other words, those nodes that change between two computational results are recognized. These nodes must be assigned or animated, for example, taking into account mathematical distances, such as Euclidean distances, between the respective nodes. These non-identical nodes may also be referred to as changing nodes.

For example, the case of mismatch or missing identity occurs when a node is present in the first lattice structure but is no longer present in the second lattice structure, or has disappeared or dropped. As a result, fewer nodes are present in the second lattice structure. The number of nodes decreases over time. This is a drop of the node, which means that the node disappears and can also be deleted.

Further, for example, a node is not present in the first lattice structure but exists in the second lattice structure. As a result, more nodes are present in the second lattice structure. The number of nodes increases over time. Accordingly, it is an addition of the node.

Furthermore, for example, the number of nodes may remain the same or unchanged. The nodes are not identical but similar. For example, the nodes are spatially not in the exact same location, but still arranged close together.

The determination according to the invention of the subsequent data record avoids the disadvantageous jumps, pauses or still images from the prior art between the iterations, computation results and ultimately individual images. The animation of the individual images is advantageously continuous and continuous.

In one embodiment, the node is designed in a multidimensional manner, comprising a position and / or at least one property. Accordingly, the node may also be considered as a multidimensional vector (X, Y, Z, T), where X, Y, Z indicate the position of the node and T the property. The property is for example the temperature. As a result, the parameters for the nodes of the at least two grid structures of the preceding data sets can be set as desired and taken into account in the comparison and the distance calculation.

In a further embodiment, the comparing comprises determining respective distances between the respective nodes of the one grid structure of the at least two grid structures and the respective nodes of the other grid structure of the at least two grid structures. Accordingly, the distances or distances are calculated. For example, the spatial or mathematical distances between the positions X, Y, Z of the nodes can be determined. Those nodes with the smallest spatial distance - ie the nearest neighbors - are determined. Alternatively or additionally, those nodes with similar temperature values can be determined, etc. As a result, advantageously, the difference or distance calculation can be adapted flexibly with regard to the user requirements.

In a further embodiment, the distance is a mathematical distance, in particular a Euclidean distance or an L1 distance.

In a further embodiment, the two nodes match if the distance between the two nodes is low or zero. Accordingly, there is a match or identity if the distances or distances between two nodes are close to zero or zero. For example, the distance is small if the nodes spatially occupy the same position in the grid structure or are arranged very close together. As a result, these nodes can be efficiently taken over into the subsequent data record without any further calculation steps or animation.

In a further embodiment, in the event that two nodes do not match, the node of the one grid structure is assigned to a corresponding node taking into account the smallest distance and the assigned node is taken over into the subsequent data record by means of an interpolation or a low-pass filter. Accordingly, advantageously only the changing nodes, for example new or dropping nodes in a grid structure, require this further step or animation.

In a further embodiment, the method further comprises:

Representing the at least two preceding records and / or the subsequent record as frames in an animation in real time.

Accordingly, the subsequent data set is determined for a real-time application. Different real-time applications are possible. For example, a real-time application is a 3D animation or illustration. The data sets are processed into corresponding individual images and displayed to the user of VR glasses in real time. This animation is advantageously up-to-date, continuous and continuous.

The invention further relates to a computer program product with a computer program which has means for carrying out the method according to the invention when the computer program is executed on a program-controlled device.

A computer program product, such as a computer program means may, for example, be used as a storage medium, e.g. Memory card, USB stick, CD-ROM, DVD, or even in the form of a downloadable file provided by a server in a network or delivered. This can be done, for example, in a wireless communication network by the transmission of a corresponding file with the computer program product or the computer program means. As a program-controlled device is in particular a control device, such as an industrial control PC or a programmable logic controller or a programmable logic controller, short PLC, or a microprocessor for a smart card or the like in question.

list of figures

• 1 zeigt ein Ablaufdiagramm zum Bestimmen mindestens eines nachfolgenden Datensatzes für eine Echtzeitanwendung gemäß der Erfindung;
• 2 zeigt einen beispielhaften Vergleich zweier vorangegangener Datensätze gemäß einer Ausführungsform der Erfindung; und;
• 3 zeigt ein Verfahren zum Verarbeiten einer Abfolge mit Iterationen in eine Animation mit Einzelbildern und deren nachteiligen Sprünge nach dem Stand der Technik.

In the following detailed description, preferred embodiments of the invention will be further described with reference to the following figures.

• 1 shows a flow chart for determining at least one subsequent record for a real-time application according to the invention;
• 2 shows an exemplary comparison of two previous data sets according to an embodiment of the invention; and;
• 3 shows a method for processing a sequence with iterations into a single-frame animation and its disadvantageous jumps in the prior art.

Description of the Preferred Embodiments

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.

1 illustrates a flowchart of the method according to the invention in detail. The method is at least one subsequent record 30 intended for a real-time application. The following record 30 follows at least two previous records in time 10 . 20 , where only the previous records 10 . 20 are known. The previous records 10 . 20 are grating structures M _i-1 and M _i .

As already explained above, the grid structures of the preceding data sets can 10 . 20 as M _i with N _i nodes 12 and M _i-1 with N _i-1 nodes 22 To be defined. N corresponds to the number of nodes at the preceding times i and i-1. In other words, the iteration result at time i is defined via a grid amount M _i with N _i nodes. The next iteration result at time i + 1 is defined by the set M _{i + 1} with N _{i + 1} node. There is no relation between N _i and N _{i + 1} . Accordingly, the number of nodes at time N _{i + 1 may be} less than, equal to, or greater than N _i .

Furthermore, the animation is a factor n faster than the parallel calculation. The following applies to the sampling times: $${\text{T}}_{\text{calculation}}=\text{n}*{\text{T}}_{\text{animation}}$$

In the n-1 time steps between two animation frames, it is possible to change the illustrated lattice structure such that at time i the result represented by N _i nodes is gently animated to that by N _{i + 1} nodes, regardless of the number of nodes , The animation frames can also be interpreted as single images.

Each node has information or parameters in the form of Cartesian coordinates about the position of the node (X, Y, Z) and / or at least one property. For example, the properties may be color-coded properties such as temperature T, mechanical stresses, electric or magnetic field strengths, etc. Accordingly, the node can also be viewed from multidimensional vector (X, Y, Z, T).

The first previous record 10 is exemplary as a rectangle in 1 and has 8 nodes. Further, the second previous record 20 exemplified as a circle in 1 and also has 8 nodes. The following record 30 can also be a dataset to be determined or unknown 30 which is based on the two records 10 . 20 is determined as follows.

In a first method step, the grid structure of the first preceding data record 10 with a variety of nodes 12 and the lattice structure of the second preceding record 20 with a variety of nodes 22 determined. In addition, further preceding data records can be determined (not shown). Furthermore, in a second method step, each node of the plurality of nodes of the first grid structure M _{i-1 is} compared with each node of the plurality of nodes of the second grid structure M _i . In other words, in each case a first vector (X ₀ , Y ₀ , Z ₀ , T ₀ ) is compared with another second vector (X ₁ , Y ₁ , Z ₁ , T ₁ ).

First, those nodes are recognized which do not change between two computation results. These nodes can be used unchanged in the new illustration step without animation.

Those accounts that change are animated. In other words, in case two nodes do not match, the node of the one of the at least two mesh structures is assigned to a corresponding node. The assigned node is taken over into the following data record. The assignment can be based on a difference calculation and the transfer on an interpolation or a low-pass filter.

wobei y den gefilterten Wert darstellt. Die wählbare Konstante a₁ hängt von der Zeitkonstante des Filters ab: $$\text{Zeitkonstante}=-{\text{T}}_{\text{Animation}}\text{/ln}\left(\text{a}1\right)\sim 1\text{/}\left(1-{\text{a}}_1\right)$$

Instead of making a sudden change in the property between two calculation steps, the change in the property is filtered using discrete time steps using a low-pass filter: $${\text{y}}_{\text{i}+1}={\text{x}}_{\text{j}\text{i}+1}-{\text{a}}_1{\text{y}}_{\text{i}}$$

where y represents the filtered value. The selectable constant a ₁ depends on the time constant of the filter: $$\text{time constant}=-{\text{T}}_{\text{animation}}\text{/ ln}\left(\text{a}1\right)~1\text{/}\left(1-{\text{a}}_1\right)$$

Since, due to the sampling, a discrete filter having a sampling time T _animation is used, according to the known Nyquist-Shannon theorem, only one signal with a time step T _animation / 2 without aliasing can be represented. That is, a minimum lower limit for the time constant of the discrete filter is T _animation / 2, thus holds $$\text{time constant}>{\text{T}}_{\text{animation}}\text{/}2$$

It is known that each filter produces a certain time delay of the filtered signal (group delay). So that the result of the filtering between two illustration steps (in a duration T _animation ) converges almost completely, the time constant must not be too high either. There is no theoretical upper limit here. The results depend on the user's perception. The time constant should also be smaller than the time step that is necessary for the calculation. Otherwise, the effect of filtering contradicts itself $$\text{time constant}<{\text{T}}_{\text{calculation}}$$

Thus it is defined for each node of the last frame which position it should take until the next step. This results in the transition of its transformation according to the next calculation result. The animation is again step by step. Accordingly, only Δi + 1 / n nodes are changed in each illustration step.

This animates the result of an iteration step by step, in which only part of the nodes to be modified is animated in each animation frame (every T _animation seconds) between two computation results. Furthermore, advantageously reduces the effect of a sudden change in the animation. For this it is necessary to start the animation only after the existence of the first iteration result.

Advantageously, an animation without jumps.

2 shows a first grating structure M _{i-1 of} a previous record 10 with 6 nodes and a plurality of possible grid structures of a second grid structure M _i a preceding data set 20 , These two lattice structures M _i-1 and M _i are compared with each other and the difference between the nodal properties is calculated.

For example, the second grid structure M _{i has} the same number of nodes with the same node properties. The first and second grating structures M _i-1 and M _i have 6 identical nodes. These can be used without animation in the grid structure M _{i + 1 of} the subsequent data set 30 be taken over.

Alternatively, however, the second grid structure M _{i may} also have fewer, more or different nodes, the other nodes differing in node properties (here position). The grating structure M _i has accordingly in 2 For example, 5 instead of 6 nodes, 7 instead of 6 nodes or 6 nodes, with a node in a node property is different. The one changing node can not be without animation in the grid structure of the subsequent record 30 but there is an animation, as explained in detail above.

Claims (8)

A method for determining at least one subsequent record (30) for a real-time application, comprising: a. Determining at least two previous records (10, 20), wherein b. the at least two data sets (10, 20) each have a grid structure with a multiplicity of nodes (12, 22), c. Comparing the respective nodes (12) of the one grid structure of the at least two grid structures with the respective nodes (22) of the other grid structure of the at least two grid structures, if two nodes (12, 22) match, the identical node (12, 22, 32) is taken over into the subsequent data record, if two nodes (12, 22) do not match, the node of the one grid structure of the at least two grid structures is assigned to a corresponding node (32) and the assigned node (32) is taken over into the subsequent data record (30). Method according to Claim 1 wherein the node (12, 22) is formed multidimensional, having a position and / or at least one property (T). Method according to Claim 1 or 2 wherein the comparing comprises determining respective distances between the respective nodes (12) of the one grid structure of the at least two grid structures and the respective nodes (22) of the other grid structure of the at least two grid structures. Method according to Claim 3 , wherein the distance is a mathematical distance (D), in particular a Euclidean distance or an L1 distance. The method of any one of the preceding claims, wherein the two nodes (12, 22) coincide when the distance between the two nodes (12, 22) is low or zero. Method according to one of the preceding claims, wherein in the event that two nodes do not match, the node of a lattice structure is assigned to a corresponding node taking into account the smallest distance and the assigned node is adopted by means of an interpolation or a low-pass filter in the subsequent data set. The method of any one of the preceding claims, the method further comprising: Representing the at least two preceding records (10, 20) and / or the subsequent record (30) as frames in an animation in real time. Computer program product with a computer program, the means for carrying out the method according to one of Claims 1 to 7 when the computer program is executed on a program-controlled device.

Images