AI News, New interactive machine learning tool makes car designs more aerodynamic

New interactive machine learning tool makes car designs more aerodynamic

'We want people to be able to design objects interactively, and therefore we work together to develop data-driven methods,' he adds.

So far, it has been extremely challenging to apply machine learning to the problem of modeling flow fields around objects because of the restrictive requirements of the method.

Two objects that look very similar to a person might therefore appear very different to a computer, as they are represented by a different mesh, and the machine would therefore be unable to transfer the information about the one to the other.

A model starts with a small number of large cubes which are then refined and split up in smaller ones following a well-defined procedure.

If represented in this way, objects with similar shapes will have a similar data structure that machine learning methods can handle and compare.

New interactive machine learning tool makes car designs more aerodynamic

When engineers or designers want to test the aerodynamic properties of the newly designed shape of a car, airplane, or other object, they would normally model the flow of air around the object by having a computer solve a complex set of equations—a procedure that usually takes hours or even an entire day.

Their method, which is the first to use machine learning to model flow around continuously editable 3D objects, will be presented at this year’s prestigious SIGGRAPH conference in Vancouver, where IST Austria researchers are involved in a total of five presentations.

So far, it has been extremely challenging to apply machine learning to the problem of modeling flow fields around objects because of the restrictive requirements of the method.

Two objects that look very similar to a person might therefore appear very different to a computer, as they are represented by a different mesh, and the machine would therefore be unable to transfer the information about the one to the other.

A model starts with a small number of large cubes which are then refined and split up in smaller ones following a well-defined procedure.

A machine learning methodology to analyze 3D digital models of cultural heritage objects

Thanks to recent advances in scanning technologies there has been an increase in the number of methods developed for digitizing cultural heritage objects.

In this paper we present some results of an ongoing project that applies machine learning and computer vision techniques for recognizing, retrieving and classifying cultural heritage objects in an automatic way (Jiménez-Badillo, et al.

Our implementation allows analyzing pairs of objects whose shapes represent the canonical extremes of a continuum, that is, objects that belong to two different “styles” within a cultural tradition.

The last condition guarantees that the deformation includes the previous knowledge of the user in terms of which forms are acceptable for the deformation, because it makes no sense to transform a face mask into an airplane, for example.

This includes masks belonging to several well-defined styles, but it also includes many others that cannot be clearly positioned within a specific class because they share features of two or more canonic styles (figure 2).

This has attracted the attention of many specialists and during the last three decades these items have been the subject of intense debate for two main reasons: First, the 162 masks were located in 14 Aztec offerings dating from 1390 to 1469 A.C., yet they do not show typical “Aztec” features.

This leads to the question: Did the Aztecs collected “antique” objects to re-use them in their own offerings?, or the Guerrero/Mezcala styles survived till the late Postclassic period and therefore the offering objects were produced during Aztecs times?

Previous studies have tried to solve some of these questions by analyzing object shapes with clustering methods (Olmedo and González, 1986 , González and Olmedo, 1990, Jiménez-Badillo and Ruiz-Correa, 2017), but we believe that the application of morphing algorithms could produce a more objective assessment to solve the problem of style attribution in this and other archaeological collections.

Our application takes examples of two canonical styles and applies the deformation algorithm in order to produce a hundred virtual 3D models whose shapes go from one to the other extreme (figure 3).

During the presentation we demonstrate a piece of software that implements the morphing algorithm and show, in a visual way, which parts of a 3D model suffer more deformation while transitioning from style “A” to “B” (figure 4).

As this is a work in progress, we are interested in receiving feedback from the audience about the relevance of our tools to resolve similar or new archaeological questions and welcome collaboration with other research projects willing to try this generic software for new applications.

3D Aerodynamic Modeling Derived from Machine Learning

The development of land, air, and sea vehicles with low drag and good stability has benefited greatly from the huge strides made in Computational Fluid Dynamics (CFD).

They have developed a method using machine learning that “learns” to mode flow around three-dimensional objects, making streamlines and parameters like drag coefficient available in real time.

'We want people to be able to design objects interactively, and therefore we work together to develop data-driven methods.'  Machine Learning in Training The technique the pair developed involves “training” the machine learning program on the converged CFD data for a variety of shapes and vehicle designs that are representative of typical vehicles.

If represented in this way, objects with similar shapes will have a similar data structure that machine learning methods can handle and compare.” Aside from the huge time savings, the method described allows modifications and shape changes to be made in real time by interactively pulling and pushing the polycubes.

The results show similar errors (approximately 3.4% in drag coefficient) as do other CFD techniques, which is consistent with the error expected when various wind tunnels are compared to one another using similar conditions.

One reason for the high level of accuracy comes directly from machine learning. 'When simulations are made in the classical way, the results for each tested shape are eventually thrown away after the computation.

General Automatic Human Shape and Motion Capture Using Volumetric Contour Cues - ECCV 2016

H. Rhodin, N. Roberitini, D. Casas, C. Richardt H-P. Seidel and C. Theobalt, General Automatic Human Shape and Motion Capture Using Volumetric Contour ...

Lecture 11 | Detection and Segmentation

In Lecture 11 we move beyond image classification, and show how convolutional networks can be applied to other core computer vision tasks. We show how ...

Designing For Movement: Evaluating Computational Models using LMA Effort Qualities

Full Title: Designing For Movement: Evaluating Computational Models using LMA Effort Qualities Authors: Diego Silang Maranan, Sarah Fdili Alaoui, Thecla ...

Riemannian manifolds, kernels and learning

I will talk about recent results from a number of people in the group on Riemannian manifolds in computer vision. In many Vision problems Riemannian ...

[PURDUE MLSS] Using Heat for Shape Understanding and Retrieval by Karthik Ramani

Using Heat for Shape Understanding and Retrieval 3D mesh segmentation is a fundamental low-level task with applications in areas as diverse as computer ...

Classifying Handwritten Digits with TF.Learn - Machine Learning Recipes #7

Last time we wrote an image classifier using TensorFlow for Poets. This time, we'll write a basic one using TF.Learn. To make it easier for you to try this out, ...

Geometric Deep Learning | Michael Bronstein || Radcliffe Institute

As part of the 2017–2018 Fellows' Presentation Series at the Radcliffe Institute for Advanced Study, Michael Bronstein RI '18 discusses the past, present, and ...

Interactive Wood Combustion for Botanical Tree Models (SIGGRAPH Asia 2017)

Interactive Wood Combustion for Botanical Tree Models Sören Pirk1 Michał Jarząbek2 Torsten Hädrich3 Dominik L. Michels3 Wojciech Palubicki2 1Stanford ...

Shape from Metric (SIGGRAPH 2018)

This is the supplementary video for the paper "Shape from Metric," where we develop theory and an algorithm for the discrete isometric immersion problem: ...

A.I. Experiments: Visualizing High-Dimensional Space

Check out to learn more. This experiment helps visualize what's happening in machine learning. It allows coders to see and explore ..