NeuralShape: the first deep-learning based software to improve the engineer ‘work.

Informations projet

NeuralShape

N° de convention de subvention: 854609

DOI

10.3030/854609

Projet clôturé

Date de signature de la CE 8 Février 2019

Date de début 1 Mars 2019

Date de fin 30 Juin 2019

Financé au titre de

• INDUSTRIAL LEADERSHIP - Innovation In SMEs

Coût total

€ 71 429,00

Contribution de l’UE

€ 50 000,00

Coordonné par

NEURAL CONCEPT SARL
Switzerland

Periodic Reporting for period 1 - NeuralShape (NeuralShape: the first deep-learning based software to improve the engineer ‘work.)

Période du rapport: 2019-03-01 au 2019-06-30

Résumé du contexte et des objectifs généraux du projet

Deep Learning (DL) algorithms have revolutionized how images, sounds and natural language are processed. Unfortunately, such techniques have only started making inroads into the domain of CAE. Neural Concept have developed NeuralShape: a patent protected AI assistant for engineers using CAD/CFD software. Our DL based SW allows to speed up R&D cycles, enhance product performance and reduce computational costs. Our tools and methods have already successfully solved several real-world problems, have beaten industry benchmarks and have shown promising results for a wide range of applications. The objective of this project is the industrial adaptation and scaling up of our innovative DL system.

Travail effectué depuis le début du projet jusqu’à la fin de la période considérée dans le rapport et principaux résultats atteints jusqu’à présent

During the FS we precisely designed the technical activities required to reach a TRL 8 through the optimization of our algorithms and our user interface as well as through validation activities carried out with industry leaders. A full market research was carried out identifying: target clients, potential customers, addressable market and competitors. A prior-art analysis was carried out concluding that we have Freedom-to-Operate in our target countries. Moreover, we established our commercialization strategy consisting on addressing engineering firms through direct sales and via online SaaS cloud interface. Additionally, we will rely on partners to distribute our solution as an integrated plugin within their SW platforms. We intend to scale-up our business first in Switzerland, UK, Germany and France.

Progrès au-delà de l’état des connaissances et impact potentiel prévu (y compris l’impact socio-économique et les conséquences sociétales plus larges du projet jusqu’à présent)

NeuralShape is the first DL system that has a notion of its own body-shape. It is agnostic to the optimization parameters as it directly works with the mesh representation of the design and a single predictor can be trained with a large amount of data while becoming more powerful over time. It can also leverage historical data from different sources without requiring designs to share an underlying parametrization. The main expected impact from NeuralShape is a reduction of design development times and the selection of optimal designs under rationally defined criteria. This translates into higher accuracy, time saving and reduction of staff requirements for CAE, which, on a broader societal scale, translate into reduced energy and resource consumption from the design process and from optimally designed vehicles, aircrafts and ships.

Neural Shape can optimize a drone design automatically in a few iterations