Erik Herrmann

Deutsches Forschungszentrum für Künstliche Intelligenz | DFKI · Agents and Simulated Reality

For years, the manufacturing industry has been investing substantial amounts of research and development work for the implementation of hybrid teams of human workers and robotic units. The composition of hybrid teams requires an optimal coordination of individual players with fundamentally different characteristics and skills. In this paper, we pre...

We propose a novel approach to create generative models for distinctive styles of locomotion for humanoid characters. Our approach only requires a single or a few style examples and a neutral motion database. We are inspired by the observation that human styles can be easily distinguished from a few examples. However, learning a generative model fo...

The simulation of humanoid avatars is relevant for a multitude of applications, such as movies, games, simulations for autonomous vehicles, virtual avatars and many more. In order to achieve the simulation of realistic and believable characters, it is important to simulate motion with the natural motion style matching the character’s characteristic...

To enable the physically correct simulation of the interaction of a 3D character with its environment the internal joint forces of a physical model of the character need to be estimated. Recently, derivative-free sampling-based optimization methods, which treat the objective function as a black box, have shown great results for finding control sign...

Human motion capture data has been widely used in data-driven character animation. In order to generate realistic, natural-looking motions, most data-driven approaches require considerable efforts of pre-processing, including motion segmentation and annotation. Existing (semi-) automatic solutions either require hand-crafted features for motion seg...

Figure 1: Our dilated temporal fully-convolutional neural network (DTFCN) for motion capture segmentation. The initial layer consists of a traditional 2D convolutional layer. The next layers are 1D temporal acausal convolutions with dilation. The final dilated conv layer uses a normalizing ReLU activation function. Finally, we add a Softmax layer a...

Semantic segmentation of motion capture sequences plays a key part in many data-driven motion synthesis frameworks. It is a preprocessing step in which long recordings of motion capture sequences are partitioned into smaller segments. Afterwards, additional methods like statistical modeling can be applied to each group of structurally-similar segme...

Statistical model-based motion synthesis enables the constrained generation of natural human motions. In this paper, we present a combination of statistical model-based motion synthesis with a motion retrieval method. We fit a statistical motion model to the low-dimensional latent space obtained by applying principal component analysis to a functio...

For ergonomic assessment of manual assembly tasks, digital simulation has received increasing attention due to its efficiency compared to physical prototypes. One of the crucial parts of digital simulation is an accurate animation of the digital human model (DHM). Current digital simulation tools such as Delmia V5 require interactive manual editing...

Many of the existing data-driven human motion synthesis methods rely on statistical modeling of motion capture data. Motion capture data is a high dimensional time-series data, therefore, it is usually required to construct an expressive latent space through dimensionality reduction methods in order to reduce the computational costs of modeling suc...

This paper describes part of a novel view of planning the assembly of cars at the shop floor, which is currently being explored in the EU-funded project INTERACT, in which 3D worker simulations are automatically generated from textual descriptions. Under this view, all planning is carried out virtually, thereby interactively exploiting the workers’...