Computer-aided manufacturing a customised product

The present invention relates to a method, an apparatus, a system, and a computer program product for manufacturing a customised product. Further the invention relates to a cus tomised product manufactured according to the said method.

Manufacturing customised products taking user-specific re quirements and/or preferences into account, usually requires an individualized design or composition, based on which the product can be manufactured. Therefore, customised products are typically manufactured in small numbers as the customisa tion process is traditionally time-consuming and/or costly.

The design process of a product is usually split into at least two separate processes. The aesthetical shaping, e.g., forming the general shape or defining a material of clothing, is performed which typically depends on the creativity and/or preferences and/or perception of a designer. This process is often performed manually and/or requires a human designer in the loop to adjust and/or optimize the design. Therefore, such a design process is not suited for mass customisation. Additionally, the functional and/or physical requirements are determined and/or designed, e.g. the insulation performance of the clothing.

Computer-aided design generation and optimization of a prod uct design, usually for technical applications, are usually based on mathematical algorithms and therefore focus on phys ical design variables, as aesthetical or subjective variables are usually difficult to quantify or to objectively measure. Thus, one of the limitations of current computer-aided design methods, e.g. for mass production of a product, can be their focus on physical quantities. It is therefore an objective of the present invention to fa cilitate manufacturing of customised products.

The object of the invention is achieved by the features of the independent claims. The dependent claims contain further developments of the invention.

The invention provides according to the first aspect a method for manufacturing a customised product, comprising the method steps :

(a) receiving a base model for the customised product, where in a first parameter of the base model is set,

(b) receiving user-specific data and deducing a constraint for base model parameters thereof,

(c) generating a customised model by adjusting a second pa rameter of the base model in accordance with the constraint,

(d) receiving a computer-aided prediction of the user' s use of the customised product according to the customised model,

(e) computing a first performance indicator by means of a computer-aided evaluation of the user' s perception to the computer-aided prediction,

(f) iteratively optimizing the first performance indicator by means of an optimization routine, wherein the second and/or a third parameter of the base model are adjusted and the method steps (c) to (e) are repeated,

(g) outputting an optimized customised model for manufactur ing the customised product

and

(h) manufacturing the customised product according to the op timized customised model.

If not indicated differently the terms "calculate", "per form", "computer-implemented", "compute", "determine", "gen erate", "configure", "reconstruct", and the like, preferably are related to acts and/or processes and/or steps which change and/or generate data, wherein data can particularly be presented as physical data, and which can be performed by a computer or processor. The term "computer" can be interpreted broadly and can be a personal computer, server, pocket-PC- device, mobile computing device, a communication device which can process data, or a processor such as a central processing unit (CPU) or microprocessor.

An important advantage of the present invention is the inte grated optimization of a model for a customised product tak ing perceptible properties into account. Hence, the prefer ences and/or taste of a user can automatically and objective ly be considered without a dedicated user input. Furthermore, the customised model is automatically individualized by tak ing the user-specific data into account, e.g., fitting gar ment to the user's body.

A customised product can for example be a manufactured prod uct. In particular, the product can be a consumer product, preferably an individualized product, such as clothing or a wearable item, which is designed and manufactured such that it meets preferably all user requirements.

The base model and the customised model preferably comprise parameters, i.e., both are parametrizable . At least one first parameter of the base model is predefined to limit the param eter space. One or more parameters can be adjusted during the optimization process. The base model or the customised model can be considered as a base design or customised design, re spectively. The base model and/or the customised model can be provided as a data structure which can be loaded in, pro cessed and/or exported by a computer.

The proposed method enables mass production of customised products, wherein each customised product corresponds to an optimized customised model. Therefore, such a set of custom ised products can comprise a multitude of parameter combina tions, such that each resulting product can preferably be unique as it is optimized according to a user's preference and requirements. According to an advantageous embodiment of the method, a sec ond performance indicator can be computed by means of a com puter-aided evaluation of a physical performance of the cus tomised product according to the customised model and the first performance indicator and/or the second performance in dicator can be optimized by means of the optimization rou tine .

It is an advantage of the proposed method that both, the physical and the perceptible properties of a model for a cus tomised product can be considered in the optimization pro cess. Furthermore, at least one customised model can be pro vided, i.e., the optimization routine can provide for example a Pareto front or Pareto optimum of parameter combinations resulting in a variety of optimized customised models.

The second performance indicator can preferably be obtained for a specific function and/or physical behavior or proper ties of the product, such as insulation or material, and/or it can be determined with respect to a specified performance threshold. At least one optimized customised model is deter mined by iteratively optimizing the first performance indica tor and/or the second performance indicator. The generated customised model is evaluated and, by means of an optimiza tion routine, the first performance indicator and/or the sec ond performance indicator can be maximized or minimized based on an objective function. Preferably, a predefined threshold or limit can be set for the first performance indicator and/or the second performance indicator if the optimization does not converge and/or for selecting at least one custom ised model .

The proposed combined optimization preferably enables finding a physically or functionally optimized customised model, which is tailored to specific user requirements and which is attractive to a user. Inversely, an aesthetically optimized customised model, which might not be functionally optimized can be discarded. According to a further embodiment the user-specific data can comprise measurement data obtained from a body scan of at least part of the user's body.

The body scan is preferably performed of at least that corre sponding part of the user' s body which is relevant to the customised product, e.g. scan of the feet for designing cus tomised shoes. The user-specific data can be used as boundary conditions for adjusting the base model according to user specific demands.

According to a further embodiment, the user-specific data can comprise input data provided by the user, wherein the input data specify a requirement for the customised product.

The input data can for example comprise physical, functional and/or perceptible requirements which are defined by the us er. The input data can for example be provided as a data structure comprising relevant information about user-specific requirements .

According to an advantageous embodiment of the method, the computer-aided prediction of the user' s use of the customised product according to the customised model can comprise a com puter-aided soft-body dynamics prediction for the customised product wherein the user' s use of the customised product is simulated .

For customised clothing, the customised garment can prefera bly be simulated during use, e.g., during movement of the us er. The computer-aided soft-body dynamics prediction is pref erably performed in real-time and/or in parallel to the opti mization process.

According to an advantageous embodiment, an output of the simulated customised product can be provided to the user by means of a user interface. The at least one generated customised model can for example be visualized and presented to the user via a display or sim ilar. The modelled product can for example be rendered as a computer-aided 3D model based on the customised model and presented to the user. If more than one optimized customised model is generated, each corresponding product can be simu lated and several models can be presented facilitating a se lection by the user.

According to an advantageous embodiment, the computer-aided evaluation of the user' s perception to the computer-aided prediction can comprise measuring user perception data in re sponse to the presentation of the simulated customised model by means of a perception capturing unit and deducing the first performance indicator from said measured user percep tion data.

The simulated customised product can be presented to a user by means of a user interface, such as a display. The human perception in response to the generated design, e.g., a us er' s reaction to the optical appearance of the at least one simulated product shown on a display, can be measured by means of a perception capturing unit or a brain-computer in terface, e.g., a consumer electroencephalogram (EEG) or an eye-tracking-sensor . The perception capturing unit is prefer ably configured to transform human perception into measurable perception data, i.e., to transform sensory input into com puter processable data.

Alternatively, according to a further embodiment, the comput er-aided evaluation of the user' s perception to the computer- aided prediction can comprise an execution of a first trained machine learning algorithm which is trained to evaluate the user' s perception to the user' s use of the customised product based on the parameters of the simulated customised model. Instead of measuring the user' s perception data, a first trained machine learning algorithm, such as a neural network, can be used to select a preferred generated customised model based. The first machine learning algorithm is preferably trained to select a preferred customised model corresponding to user-specific preferences, wherein the training can be based on training data comprising information about the user and/or the user's preferences and/or data of a peer group.

According to an advantageous embodiment the computer-aided evaluation of the physical performance of the customised product can comprise a computer-aided physical or functional simulation of the customised product according to the custom ised model and wherein the second performance indicator is computed from the computer-aided physical or functional simu lation .

Using a computer-aided physical simulation, reproducing phys ical properties and/or constraints of the product, a function or physical property of the customised model can be tested and/or evaluated. Preferably, the generated customised model is provided in a computer-readable format such that it can be used as an input for a computer-aided simulation. The second performance indicator can be deduced from the computer-aided simulation .

Alternatively, according to a further embodiment, the comput er-aided evaluation of the physical performance of the cus tomised product can comprise an execution of a second trained machine learning algorithm which is trained to evaluate the physical performance of the customised product based on the parameters of the customised model.

Instead of performing a computer aided physical or functional simulation of the customised product, a trained second ma chine learning algorithm, e.g. a neural network, can be exe cuted using the parameters of the according customised model as input for the second machine learning algorithm. The sec- ond machine learning algorithm can preferably be trained to select a customised model based on its parameters which has required physical or functional properties.

According to a further embodiment a weight can be allocated to the first performance indicator and/or a weight can be al located to the second performance indicator and the optimiza tion algorithm is configured to take the respective weight into account .

Preferably the user of the customised product can choose dif ferent criteria for optimization and weights of different ob jectives. Therefore, the individualization of the product can be further achieved by prioritization. A weight can for exam ple be configured as a statistical weight.

According to an advantageous embodiment the optimized custom ised model can be provided to a manufacturing unit for manu facturing the customised product by the manufacturing unit using the optimized customised model.

According to a further embodiment a label based on the opti mized first performance indicator and/or the optimized second performance indicator and/or the user-specific data can be allocated to the customised model and/or the customised prod uct .

The label can for example be incorporated in or attached to the customised product such that the product can further be identified based on the respective label, such that e.g. emu lation can be avoided. Furthermore, labeling the customised product according to the optimized first performance indica tor and/or the optimized second performance indicator and/or the user-specific data can allow verifying that a manufac tured product has been produced according to the proposed method. The label can advantageously be configured as comput er-readable and/or encoded and/or physically protected. The invention provides according to the second aspect an ap paratus for manufacturing a customised product, comprising:

(a) an interface unit configured to

- receive a base model for the customised product, wherein a first parameter of the base model is set,

- receive user-specific data and to deduce a constraint for base model parameters thereof,

(b) a model generator configured to generate a customised model by adjusting a second parameter of the base model in accordance with the constraint,

(c) a computing unit configured to

- receive a computer-aided prediction of the user' s use of the customised product according to the customised model,

- compute a first performance indicator by means of a comput er-aided evaluation of the user' s perception to the computer- aided prediction,

(d) an optimization unit configured to iteratively optimize the first performance indicator by means of an optimization algorithm, wherein a third parameter of the base model is ad justed,

and

(e) an output unit configured to output an optimized custom ised model for manufacturing the customised product.

The apparatus and/or at least one of its units can further comprise at least one processor or computer to perform the method steps according to the invention. Furthermore, at least one of the respective units can be realized by means of cloud computing.

The respective unit may be implemented in hardware and/or in software. If said unit is implemented in hardware, it may be embodied as a device, e.g. as a computer or as a processor or as a part of a system. If said unit is implemented in soft ware it may be embodied as a computer program product, as a function, as a routine, as a program code or as an executable object. The output unit preferably provides a data structure comprising the optimized customised model.

The invention provides according to the third aspect a system for manufacturing a customised product, the system comprising an apparatus according to the invention and a manufacturing unit configured to manufacture the customised product.

Preferably, the manufacturing unit is configured to receive the optimized customised model and to manufacture the custom ised product according to that customised model. Preferably the customised model is provided in a data format which is processible by the manufacturing unit.

The invention further comprises a computer program product directly loadable into the internal memory of a digital com puter, comprising software code portions for performing the steps of the said method when said product is run on a com puter .

A computer program product, such as a computer program means, may be embodied as a memory card, USB stick, CD-ROM, DVD or as a file which may be downloaded from a server in a network. For example, such a file may be provided by transferring the file comprising the computer program product from a wireless communication network.

Further the invention relates to a customised product which is manufactured by performing a method according to the in vention and which is labeled based on the optimized first performance indicator and/or the optimized second performance indicator and/or the user-specific data.

The labeling of such manufactured customised product allows for example unique identification and/or traceability of the customised product. The invention will be explained in more detail by reference to the accompanying figures.

Fig. 1 shows a flow chart including method steps involved in an embodiment of a method for manufacturing a customised product according to the invention;

Fig . 2 shows a first schematic representation of an embodi ment of the method for manufacturing a customised product according to the invention;

Fig . 3 shows a second schematic representation of an embod iment of the method for manufacturing a customised product according to the invention;

Fig. 4 shows a schematic diagram of an embodiment of an ap paratus for manufacturing a customised product ac cording to the invention; and

Fig . 5 shows a schematic diagram of an embodiment of a sys tem for manufacturing a customised product according to the invention.

Equivalent parts in the different figures are labeled with the same reference signs.

Figure 1 shows a flow chart including method steps of a meth od according to the invention for manufacturing a customised product .

The first step SI of the method involves receiving a base model for the customised product. The base model is prefera bly parametrized, wherein at least one first parameter is set. The first parameter can for example define the type, ma terial, or color of the product. By setting at least one pa rameter the configurable parameter space of the base model can be limited such that the following optimization process can be advantageously constrained, and the according compu ting time can be reduced.

The base model can be provided as a data structure in a com puter-readable format. The base model serves in particular as a starting point for the customisation of the product accord ing to specific requirements and/or preferences of a user.

Further, in the next step S2 which can also be performed in parallel or before the first step SI, user-specific data are received and at least one constraint for the base model is deduced thereof. User-specific data can for example be input data comprising requirements or preferences for the custom ised product and/or measurement data of a body scan of the user's body or part of the user's body.

In the next step S3, based on the provided user-specific da ta, at least one second parameter of the base model is ad justed such that the requirements of the user-specific data are fulfilled, and a customised model is generated. In other words, one or more first parameters of the base model are set such that the boundary conditions deduced from the user- specific data are met. Therefore, the generated customised model of the product results from the adjusted base model.

The next step S4 includes receiving a computer-aided predic tion of the user' s use of the customised product according to the generated customised model. The computer-aided prediction of the user' s use can be for example a simulation of the cus tomised product such as a soft-body dynamics prediction.

In the next step S5, a first performance indicator is comput ed based on a computer-aided evaluation of the provided com puter-aided prediction of the user' s use of the customised product. In other words, a first performance indicator quan tifying the user's perception towards the product, is comput ed by means of a computer-aided evaluation of the computer- aided prediction of the used customised product. Afterwards, in the next step S6, the first performance indi cator is optimized using an optimization routine, wherein the steps S3 to S5 are iteratively repeated until an optimum of the first performance indicator is found, corresponding to an optimized customised model. During optimization of the first performance indicator at least one second and/or third param eter of the base model is adjusted such that a different cus tomised model is generated which is subsequently evaluated as described .

When an optimum is reached, the optimized customised model is outputted, step S7. An optimum can for example also be reached when the deduced first key performance indicator passes a predefined threshold. The optimized customised model can for example be exported in a computer-readable format and can such be submitted to a manufacturing unit to manufacture the customised product. The optimized model can for example be provided as a data structure and/or template, such as cuts in case of customised cloths.

Figure 2 shows a first schematic representation of an embodi ment of the method for manufacturing a customised product ac cording to the invention. The manufacturing of a customised garment is shown. First, a base model 10 of the garment is provided, see step SI. The base model is preferably para metrized, wherein at least one first parameter is set. For example, the first parameter specifies the type of the gar ment, such as "dress", "shoes" or "hat". Further parameters of the base model are preferably not defined and are deter mined during the next steps. Preferably, the at least one first parameter cannot be further adjusted during the next steps in order to constrain the parameter space.

In addition, user-specific data 20 are provided and con straints for the base model are deduced thereof, step S2. The user-specific data 20 can for example be received from a body scan of the user's body. The user-specific body proportions are captured and constraints, such as size of the dress, are deduced thereof. Alternatively, or in addition, user-specific requirements, such as type of material of the garment or in sulation properties, can further be provided as user-specific input data.

Based on the at least one deduced constraint, the base model of the dress is parametrized in accordance with the user- specific data 20, i.e., the size of the dress is adjusted such that it fits to the user' s body, and a resulting custom ised model 30 is generated and outputted.

In order to quantify the user's perception towards the cus tomised model 30, a computer-aided simulation of the custom ised product according to the generated customised model 30 is performed, step S4a. Preferably, a soft-body dynamics pre diction of the product during use is computed and the result ing simulated customised product is visualized 301. The simu lated dress as it appears during use can for example be pre sented 301 to the user by means of a machine-user-interface, e.g., a display or a virtual-reality device. The user's per ception to the prediction of the customised product is evalu ated, step S4b, using for example a perception capturing de vice, e.g., an eye tracking sensor or a consumer-EEG . Sensory input data 40 of the user USR is measured by means of the perception capturing unit and a first performance indicator KPI1 is computed based on said sensory input data 40. The first performance indicator KPI1 can for example specify and/or evaluate the amount or rank of approval and/or the re sponse to the presented customised model by the user.

In addition, a physical or functional computer-aided simula tion can be performed based on the generated customised model 30 in order to evaluate the physical or functional perfor mance of the customised product, S4c. A second performance indicator KPI2 specifying the physical or functional perfor mance of the respective customised product is computed based on the computer-aided simulation. The second performance in- dicator KPI2 can for example specify and/or evaluate the physical or functional properties of the customised model, e.g., the material properties of the dress as a function of temperature .

The first performance indicator KPI1 and the second perfor mance indicator KPI2 are outputted and transferred to an op timization routine OPT. By adjusting at least one second or at least one third parameter of the base model 10, generating an alternative customised model, and computing an updated first performance indicator KPI1 and second performance indi cator KPI2, the customised model is iteratively optimized,

S6. Therefore, the customised model is gradually adjusted and improved such that it meets the user' s requirements and pref erences. The optimization can for example result in a Pareto optimum 60, resulting in several optimized customised models.

The optimization of the first and second performance indica tor KPI1, KPI2 can further be prioritized by introducing a specific weight to the first performance indicator KPI1 or second performance indicator KPI2, respectively. For example, a user can provide a prioritization or user-specific data comprise weighting information and the weights are set ac cordingly. The weighting of the performance indicators KPI1, KPI2 can further be used in the optimization routine.

At least one optimized customised model is outputted S7. It can for example be provided in a computer-readable data for mat and transferred to a manufacturing unit to manufacture the customised dress 50 according to the optimized customised model .

The manufactured dress 50 and/or the according customised model can further be labeled based on the optimized first performance indicator KPI1 and/or the optimized second per formance indicator KPI2 and/or the user-specific data 20. A label can for example be incorporated into and/or allocated to the garment. Furthermore, the label can be encoded and/or physically protected to for example ensure verification of authenticity of the customised product.

Figure 3 shows another schematic representation of an embodi ment of the inventive method for manufacturing a customised garment. Based on the provided base model 10 and the user- specific data 20 a customised model is generated, S3. The customised model is subsequently simulated S4a to predict the user's use of the customised product. The user's perception to the user' s use of the customised product is evaluated us ing a trained first machine learning algorithm ML1. The first machine learning algorithm ML1 can for example be trained us ing training data comprising information about the user, the user's preferences and/or information about the user's peer groups. Preferably the first machine learning algorithm ML1 is trained such that a first performance indicator KPI1 can be deduced from the parameters of the simulated customised model .

By executing the first trained machine learning algorithm ML1, the simulated customised model is evaluated, resulting in a performance indicator KPI1.

In addition, the second performance indicator KPI2 can be de termined using a second trained machine learning algorithm ML2 which is preferably trained to evaluate the physical or functional performance of the customised model based on the respective parameters of the respective customised model. The second machine learning algorithm can for example be trained based on a training set comprising at least information about physical and/or functional properties of product models ac cording to respective model parameters.

The first machine learning algorithm ML1 and/or the second machine learning algorithm ML2 can for example be a neural network . The first and second performance indicator KPI1, KPI2 can it eratively be optimized, S6, by means of an optimization rou tine OPT, wherein the parameters of the customised model are iteratively adjusted until an optimum is reached. If the re sulting optimized customised models form a Pareto front 60, all or a selection of these optimized customised models can be outputted S7 for manufacturing a customised product.

The method steps according to the invention shown in Fig. 4 can advantageously be combined or replaced with method steps, in particular the method steps comprising the evaluation of the respective generated customised model, shown in Fig. 2.

Figure 4 shows a schematic diagram of an embodiment of an ap paratus 100 for manufacturing a customised product according to the invention. The apparatus 100 comprises an interface unit 101, a model generator 102, a computing unit 103, an op timization unit 104, and an output unit 105. The interface unit 101 is configured to receive a base model and user- specific data. The model generator 102 is configured to gen erate a customised model by adjusting at least a second pa rameter of the base model in accordance with the constraint given by the user-specific data. Furthermore, during optimi zation, the model generator 102 is configured to generate a customised model based on adjusted second and/or third param eter of the base model, wherein the adjusted parameter values are preferably provided by the optimization unit 104. The computing unit 103 is configured to receive a computer-aided prediction of the user' s use of the customised product ac cording to the customised model and to compute a first per formance indicator by means of a computer-aided evaluation of the user's perception to the computer-aided prediction. The computing unit 103 can for example further perform a physical or functional simulation of the customised product and deduce a second performance indicator from said simulation. Further more, the computing unit 103 can be coupled to a perception capturing unit (not shown) in order to perform a computer- aided evaluation of the user' s perception to the computer- aided prediction comprising the measurement of user percep tion data in response to an output of the simulated custom ised model. The optimization unit 104 is configured to itera tively optimize the first performance indicator, and if com- puted, also the second performance indicator, to provide at least one optimized customised model. The output unit 105 is configured to output the optimized model.

Figure 5 shows a schematic diagram of an embodiment of a sys- tern 200 for manufacturing a customised product 50 according to the invention. The system 200 can comprise an apparatus 100 as shown in Figure 4 and a manufacturing unit 300 which is configured to manufacture a customised product 50 accord ing to an optimized customised model provided by the appa- ratus 100.

Although the present invention has been described in detail with reference to the advantageous embodiment, it is to be understood that the present invention is not limited by the disclosed examples, and that numerous additional modifica tions and variations could be made thereto by a person skilled in the art without departing from the scope of the invention .