Header logo is


2018


no image
Deep Reinforcement Learning for Event-Triggered Control

Baumann, D., Zhu, J., Martius, G., Trimpe, S.

In Proceedings of the 57th IEEE International Conference on Decision and Control, Miami, Fl, USA, December 2018 (inproceedings) Accepted

al ics

arXiv PDF [BibTex]

2018


arXiv PDF [BibTex]


no image
When do random forests fail?

Tang, C., Garreau, D., von Luxburg, U.

In Proceedings Neural Information Processing Systems, Neural Information Processing Systems (NIPS 2018) , December 2018 (inproceedings)

slt

Project Page [BibTex]

Project Page [BibTex]


Thumb xl pac gp img2
Learning Gaussian Processes by Minimizing PAC-Bayesian Generalization Bounds

Reeb, D., Doerr, A., Gerwinn, S., Rakitsch, B.

In Proceedings Neural Information Processing Systems, Neural Information Processing Systems (NIPS) , December 2018 (inproceedings)

Abstract
Gaussian Processes (GPs) are a generic modelling tool for supervised learning. While they have been successfully applied on large datasets, their use in safety critical applications is hindered by the lack of good performance guarantees. To this end, we propose a method to learn GPs and their sparse approximations by directly optimizing a PAC-Bayesian bound on their generalization performance, instead of maximizing the marginal likelihood. Besides its theoretical appeal, we find in our evaluation that our learning method is robust and yields significantly better generalization guarantees than other common GP approaches on several regression benchmark datasets.

ics

[BibTex]

[BibTex]


Thumb xl unbenannte pr%c3%a4sentation 1
Efficient Encoding of Dynamical Systems through Local Approximations

Solowjow, F., Mehrjou, A., Schölkopf, B., Trimpe, S.

In Proceedings of the 57th IEEE International Conference on Decision and Control (CDC), Miami, Fl, USA, December 2018 (inproceedings) Accepted

ei ics

arXiv PDF Project Page [BibTex]

arXiv PDF Project Page [BibTex]


no image
Flex-Convolution (Million-Scale Point-Cloud Learning Beyond Grid-Worlds)

Groh*, F., Wieschollek*, P., Lensch, H. P. A.

Computer Vision - 14th Asian Conference on Computer Vision (ACCV), December 2018, *equal contribution (conference) Accepted

ei

[BibTex]

[BibTex]


Thumb xl dip final
Deep Inertial Poser: Learning to Reconstruct Human Pose from Sparse Inertial Measurements in Real Time

Huang, Y., Kaufmann, M., Aksan, E., Black, M. J., Hilliges, O., Pons-Moll, G.

ACM Transactions on Graphics, (Proc. SIGGRAPH Asia), 37, pages: 185:1-185:15, ACM, November 2018, Two first authors contributed equally (article)

Abstract
We demonstrate a novel deep neural network capable of reconstructing human full body pose in real-time from 6 Inertial Measurement Units (IMUs) worn on the user's body. In doing so, we address several difficult challenges. First, the problem is severely under-constrained as multiple pose parameters produce the same IMU orientations. Second, capturing IMU data in conjunction with ground-truth poses is expensive and difficult to do in many target application scenarios (e.g., outdoors). Third, modeling temporal dependencies through non-linear optimization has proven effective in prior work but makes real-time prediction infeasible. To address this important limitation, we learn the temporal pose priors using deep learning. To learn from sufficient data, we synthesize IMU data from motion capture datasets. A bi-directional RNN architecture leverages past and future information that is available at training time. At test time, we deploy the network in a sliding window fashion, retaining real time capabilities. To evaluate our method, we recorded DIP-IMU, a dataset consisting of 10 subjects wearing 17 IMUs for validation in 64 sequences with 330,000 time instants; this constitutes the largest IMU dataset publicly available. We quantitatively evaluate our approach on multiple datasets and show results from a real-time implementation. DIP-IMU and the code are available for research purposes.

ps

data code pdf preprint video DOI [BibTex]

data code pdf preprint video DOI [BibTex]


Thumb xl universal custom complex magnetic spring design methodology
Universal Custom Complex Magnetic Spring Design Methodology

Woodward, M. A., Sitti, M.

IEEE Transactions on Magnetics, 54(1):1-13, October 2018 (article)

Abstract
A design methodology is presented for creating custom complex magnetic springs through the design of force-displacement curves. This methodology results in a magnet configuration, which will produce a desired force-displacement relationship. Initially, the problem is formulated and solved as a system of linear equations. Then, given the limited likelihood of a single solution being feasibly manufactured, key parameters of the solution are extracted and varied to create a family of solutions. Finally, these solutions are refined using numerical optimization. Given the properties of magnets, this methodology can create any well-defined function of force versus displacement and is model-independent. To demonstrate this flexibility, a number of example magnetic springs are designed; one of which, designed for use in a jumping-gliding robot's shape memory alloy actuated clutch, is manufactured and experimentally characterized. Due to the scaling of magnetic forces, the displacement region which these magnetic springs are most applicable is that of millimeters and below. However, this region is well situated for miniature robots and smart material actuators, where a tailored magnetic spring, designed to compliment a component, can enhance its performance while adding new functionality. The methodology is also expendable to variable interactions and multi-dimensional magnetic field design.

pi

DOI [BibTex]

DOI [BibTex]


Thumb xl sevillagcpr
On the Integration of Optical Flow and Action Recognition

Sevilla-Lara, L., Liao, Y., Guney, F., Jampani, V., Geiger, A., Black, M. J.

In German Conference on Pattern Recognition (GCPR), October 2018 (inproceedings)

Abstract
Most of the top performing action recognition methods use optical flow as a "black box" input. Here we take a deeper look at the combination of flow and action recognition, and investigate why optical flow is helpful, what makes a flow method good for action recognition, and how we can make it better. In particular, we investigate the impact of different flow algorithms and input transformations to better understand how these affect a state-of-the-art action recognition method. Furthermore, we fine tune two neural-network flow methods end-to-end on the most widely used action recognition dataset (UCF101). Based on these experiments, we make the following five observations: 1) optical flow is useful for action recognition because it is invariant to appearance, 2) optical flow methods are optimized to minimize end-point-error (EPE), but the EPE of current methods is not well correlated with action recognition performance, 3) for the flow methods tested, accuracy at boundaries and at small displacements is most correlated with action recognition performance, 4) training optical flow to minimize classification error instead of minimizing EPE improves recognition performance, and 5) optical flow learned for the task of action recognition differs from traditional optical flow especially inside the human body and at the boundary of the body. These observations may encourage optical flow researchers to look beyond EPE as a goal and guide action recognition researchers to seek better motion cues, leading to a tighter integration of the optical flow and action recognition communities.

am ps

arXiv [BibTex]

arXiv [BibTex]


no image
Regularizing Reinforcement Learning with State Abstraction

Akrour, R., Veiga, F., Peters, J., Neuman, G.

Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), October 2018 (conference) Accepted

ei

link (url) [BibTex]

link (url) [BibTex]


Thumb xl iros18
Towards Robust Visual Odometry with a Multi-Camera System

Liu, P., Geppert, M., Heng, L., Sattler, T., Geiger, A., Pollefeys, M.

In International Conference on Intelligent Robots and Systems (IROS) 2018, International Conference on Intelligent Robots and Systems, October 2018 (inproceedings)

Abstract
We present a visual odometry (VO) algorithm for a multi-camera system and robust operation in challenging environments. Our algorithm consists of a pose tracker and a local mapper. The tracker estimates the current pose by minimizing photometric errors between the most recent keyframe and the current frame. The mapper initializes the depths of all sampled feature points using plane-sweeping stereo. To reduce pose drift, a sliding window optimizer is used to refine poses and structure jointly. Our formulation is flexible enough to support an arbitrary number of stereo cameras. We evaluate our algorithm thoroughly on five datasets. The datasets were captured in different conditions: daytime, night-time with near-infrared (NIR) illumination and night-time without NIR illumination. Experimental results show that a multi-camera setup makes the VO more robust to challenging environments, especially night-time conditions, in which a single stereo configuration fails easily due to the lack of features.

avg

pdf [BibTex]

pdf [BibTex]


Thumb xl stco paper figure11
Probabilistic Solutions To Ordinary Differential Equations As Non-Linear Bayesian Filtering: A New Perspective

Tronarp, F., Kersting, H., Särkkä, S., Hennig, P.

ArXiv preprint 2018, arXiv:1807.09737 [stat.ME], October 2018 (article) In revision

Abstract
We formulate probabilistic numerical approximations to solutions of ordinary differential equations (ODEs) as problems in Gaussian process (GP) regression with non-linear measurement functions. This is achieved by defining the measurement sequence to consists of the observations of the difference between the derivative of the GP and the vector field evaluated at the GP---which are all identically zero at the solution of the ODE. When the GP has a state-space representation, the problem can be reduced to a Bayesian state estimation problem and all widely-used approximations to the Bayesian filtering and smoothing problems become applicable. Furthermore, all previous GP-based ODE solvers, which were formulated in terms of generating synthetic measurements of the vector field, come out as specific approximations. We derive novel solvers, both Gaussian and non-Gaussian, from the Bayesian state estimation problem posed in this paper and compare them with other probabilistic solvers in illustrative experiments.

pn

link (url) Project Page [BibTex]

link (url) Project Page [BibTex]


no image
Learning to Categorize Bug Reports with LSTM Networks

Gondaliya, K., Peters, J., Rueckert, E.

Proceedings of the 10th International Conference on Advances in System Testing and Validation Lifecycle (VALID), October 2018 (conference) Accepted

ei

link (url) [BibTex]

link (url) [BibTex]


no image
Domain Randomization for Simulation-Based Policy Optimization with Transferability Assessment

Muratore, F., Treede, F., Gienger, M., Peters, J.

Conference on Robot Learning (CoRL), October 2018 (conference) Accepted

ei

[BibTex]

[BibTex]


Thumb xl cover
Deep Neural Network-based Cooperative Visual Tracking through Multiple Micro Aerial Vehicles

Price, E., Lawless, G., Ludwig, R., Martinovic, I., Buelthoff, H. H., Black, M. J., Ahmad, A.

IEEE Robotics and Automation Letters, Robotics and Automation Letters, 3, pages: 3193-3200, IEEE, October 2018, Also accepted and presented in the 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). (article)

Abstract
Multi-camera tracking of humans and animals in outdoor environments is a relevant and challenging problem. Our approach to it involves a team of cooperating micro aerial vehicles (MAVs) with on-board cameras only. DNNs often fail at objects with small scale or far away from the camera, which are typical characteristics of a scenario with aerial robots. Thus, the core problem addressed in this paper is how to achieve on-board, online, continuous and accurate vision-based detections using DNNs for visual person tracking through MAVs. Our solution leverages cooperation among multiple MAVs and active selection of most informative regions of image. We demonstrate the efficiency of our approach through simulations with up to 16 robots and real robot experiments involving two aerial robots tracking a person, while maintaining an active perception-driven formation. ROS-based source code is provided for the benefit of the community.

ps

Published Version link (url) DOI Project Page [BibTex]

Published Version link (url) DOI Project Page [BibTex]


no image
Control of Musculoskeletal Systems using Learned Dynamics Models

Büchler, D., Calandra, R., Schölkopf, B., Peters, J.

IEEE Robotics and Automation Letters, 3(4):3161-3168, IEEE, October 2018 (article)

Abstract
Controlling musculoskeletal systems, especially robots actuated by pneumatic artificial muscles, is a challenging task due to nonlinearities, hysteresis effects, massive actuator de- lay and unobservable dependencies such as temperature. Despite such difficulties, muscular systems offer many beneficial prop- erties to achieve human-comparable performance in uncertain and fast-changing tasks. For example, muscles are backdrivable and provide variable stiffness while offering high forces to reach high accelerations. In addition, the embodied intelligence deriving from the compliance might reduce the control demands for specific tasks. In this paper, we address the problem of how to accurately control musculoskeletal robots. To address this issue, we propose to learn probabilistic forward dynamics models using Gaussian processes and, subsequently, to employ these models for control. However, Gaussian processes dynamics models cannot be set-up for our musculoskeletal robot as for traditional motor- driven robots because of unclear state composition etc. We hence empirically study and discuss in detail how to tune these approaches to complex musculoskeletal robots and their specific challenges. Moreover, we show that our model can be used to accurately control an antagonistic pair of pneumatic artificial muscles for a trajectory tracking task while considering only one- step-ahead predictions of the forward model and incorporating model uncertainty.

ei

RAL18final link (url) DOI [BibTex]

RAL18final link (url) DOI [BibTex]


Thumb xl interpolation
Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation

Wulff, J., Black, M. J.

In German Conference on Pattern Recognition (GCPR), October 2018 (inproceedings)

Abstract
The difficulty of annotating training data is a major obstacle to using CNNs for low-level tasks in video. Synthetic data often does not generalize to real videos, while unsupervised methods require heuristic n losses. Proxy tasks can overcome these issues, and start by training a network for a task for which annotation is easier or which can be trained unsupervised. The trained network is then fine-tuned for the original task using small amounts of ground truth data. Here, we investigate frame interpolation as a proxy task for optical flow. Using real movies, we train a CNN unsupervised for temporal interpolation. Such a network implicitly estimates motion, but cannot handle untextured regions. By fi ne-tuning on small amounts of ground truth flow, the network can learn to fill in homogeneous regions and compute full optical flow fi elds. Using this unsupervised pre-training, our network outperforms similar architectures that were trained supervised using synthetic optical flow.

ps

pdf arXiv [BibTex]

pdf arXiv [BibTex]


no image
Constraint-Space Projection Direct Policy Search

Akrour, R., Peters, J., Neuman, G.

14th European Workshop on Reinforcement Learning (EWRL), October 2018 (conference)

ei

link (url) [BibTex]

link (url) [BibTex]


Thumb xl coma faces
Generating 3D Faces using Convolutional Mesh Autoencoders

Ranjan, A., Bolkart, T., Sanyal, S., Black, M. J.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
Learned 3D representations of human faces are useful for computer vision problems such as 3D face tracking and reconstruction from images, as well as graphics applications such as character generation and animation. Traditional models learn a latent representation of a face using linear subspaces or higher-order tensor generalizations. Due to this linearity, they can not capture extreme deformations and non-linear expressions. To address this, we introduce a versatile model that learns a non-linear representation of a face using spectral convolutions on a mesh surface. We introduce mesh sampling operations that enable a hierarchical mesh representation that captures non-linear variations in shape and expression at multiple scales within the model. In a variational setting, our model samples diverse realistic 3D faces from a multivariate Gaussian distribution. Our training data consists of 20,466 meshes of extreme expressions captured over 12 different subjects. Despite limited training data, our trained model outperforms state-of-the-art face models with 50% lower reconstruction error, while using 75% fewer parameters. We also show that, replacing the expression space of an existing state-of-the-art face model with our autoencoder, achieves a lower reconstruction error. Our data, model and code are available at http://coma.is.tue.mpg.de/.

ps

code paper supplementary link (url) [BibTex]


Thumb xl toc image
Fast spatial scanning of 3D ultrasound fields via thermography

Melde, K., Qiu, T., Fischer, P.

Applied Physics Letters, 113(13):133503, September 2018 (article)

Abstract
We propose and demonstrate a thermographic method that allows rapid scanning of ultrasound fields in a volume to yield 3D maps of the sound intensity. A thin sound-absorbing membrane is continuously translated through a volume of interest while a thermal camera records the evolution of its surface temperature. The temperature rise is a function of the absorbed sound intensity, such that the thermal image sequence can be combined to reveal the sound intensity distribution in the traversed volume. We demonstrate the mapping of ultrasound fields, which is several orders of magnitude faster than scanning with a hydrophone. Our results are in very good agreement with theoretical simulations.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


Thumb xl ianeccv18
Learning Priors for Semantic 3D Reconstruction

Cherabier, I., Schönberger, J., Oswald, M., Pollefeys, M., Geiger, A.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
We present a novel semantic 3D reconstruction framework which embeds variational regularization into a neural network. Our network performs a fixed number of unrolled multi-scale optimization iterations with shared interaction weights. In contrast to existing variational methods for semantic 3D reconstruction, our model is end-to-end trainable and captures more complex dependencies between the semantic labels and the 3D geometry. Compared to previous learning-based approaches to 3D reconstruction, we integrate powerful long-range dependencies using variational coarse-to-fine optimization. As a result, our network architecture requires only a moderate number of parameters while keeping a high level of expressiveness which enables learning from very little data. Experiments on real and synthetic datasets demonstrate that our network achieves higher accuracy compared to a purely variational approach while at the same time requiring two orders of magnitude less iterations to converge. Moreover, our approach handles ten times more semantic class labels using the same computational resources.

avg

pdf suppmat Project Page Video [BibTex]

pdf suppmat Project Page Video [BibTex]


no image
Separating Reflection and Transmission Images in the Wild

Wieschollek, P., Gallo, O., Gu, J., Kautz, J.

European Conference on Computer Vision (ECCV), September 2018 (conference)

ei

link (url) [BibTex]

link (url) [BibTex]


Thumb xl person reid.001
Part-Aligned Bilinear Representations for Person Re-identification

Suh, Y., Wang, J., Tang, S., Mei, T., Lee, K. M.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
Comparing the appearance of corresponding body parts is essential for person re-identification. However, body parts are frequently misaligned be- tween detected boxes, due to the detection errors and the pose/viewpoint changes. In this paper, we propose a network that learns a part-aligned representation for person re-identification. Our model consists of a two-stream network, which gen- erates appearance and body part feature maps respectively, and a bilinear-pooling layer that fuses two feature maps to an image descriptor. We show that it results in a compact descriptor, where the inner product between two image descriptors is equivalent to an aggregation of the local appearance similarities of the cor- responding body parts, and thereby significantly reduces the part misalignment problem. Our approach is advantageous over other pose-guided representations by learning part descriptors optimal for person re-identification. Training the net- work does not require any part annotation on the person re-identification dataset. Instead, we simply initialize the part sub-stream using a pre-trained sub-network of an existing pose estimation network and train the whole network to minimize the re-identification loss. We validate the effectiveness of our approach by demon- strating its superiority over the state-of-the-art methods on the standard bench- mark datasets including Market-1501, CUHK03, CUHK01 and DukeMTMC, and standard video dataset MARS.

ps

pdf supplementary [BibTex]

pdf supplementary [BibTex]


Thumb xl screenshot from 2017 07 27 17 24 14
Playful: Reactive Programming for Orchestrating Robotic Behavior

Berenz, V., Schaal, S.

IEEE Robotics Automation Magazine, 25(3):49-60, September 2018 (article) In press

Abstract
For many service robots, reactivity to changes in their surroundings is a must. However, developing software suitable for dynamic environments is difficult. Existing robotic middleware allows engineers to design behavior graphs by organizing communication between components. But because these graphs are structurally inflexible, they hardly support the development of complex reactive behavior. To address this limitation, we propose Playful, a software platform that applies reactive programming to the specification of robotic behavior.

am

playful website playful_IEEE_RAM link (url) DOI [BibTex]


Thumb xl persondetect  copy
Learning Human Optical Flow

Ranjan, A., Romero, J., Black, M. J.

In 29th British Machine Vision Conference, September 2018 (inproceedings)

Abstract
The optical flow of humans is well known to be useful for the analysis of human action. Given this, we devise an optical flow algorithm specifically for human motion and show that it is superior to generic flow methods. Designing a method by hand is impractical, so we develop a new training database of image sequences with ground truth optical flow. For this we use a 3D model of the human body and motion capture data to synthesize realistic flow fields. We then train a convolutional neural network to estimate human flow fields from pairs of images. Since many applications in human motion analysis depend on speed, and we anticipate mobile applications, we base our method on SpyNet with several modifications. We demonstrate that our trained network is more accurate than a wide range of top methods on held-out test data and that it generalizes well to real image sequences. When combined with a person detector/tracker, the approach provides a full solution to the problem of 2D human flow estimation. Both the code and the dataset are available for research.

ps

video code pdf link (url) [BibTex]

video code pdf link (url) [BibTex]


Thumb xl bmvc pic
Human Motion Parsing by Hierarchical Dynamic Clustering

Zhang, Y., Tang, S., Sun, H., Neumann, H.

British Machine Vision Conference, September 2018 (conference)

Abstract
Parsing continuous human motion into meaningful segments plays an essential role in various applications. In this work, we propose a hierarchical dynamic clustering framework to derive action clusters from a sequence of local features in an unsuper- vised bottom-up manner. We systematically investigate the modules in this framework and particularly propose diverse temporal pooling schemes, in order to realize accurate temporal action localization. We demonstrate our method on two motion parsing tasks: temporal action segmentation and abnormal behavior detection. The experimental results indicate that the proposed framework is significantly more effective than the other related state-of-the-art methods on several datasets.

ps

pdf [BibTex]

pdf [BibTex]


Thumb xl nbf
Neural Body Fitting: Unifying Deep Learning and Model-Based Human Pose and Shape Estimation

(Best Student Paper Award)

Omran, M., Lassner, C., Pons-Moll, G., Gehler, P. V., Schiele, B.

3DV, September 2018 (conference)

Abstract
Direct prediction of 3D body pose and shape remains a challenge even for highly parameterized deep learning models. Mapping from the 2D image space to the prediction space is difficult: perspective ambiguities make the loss function noisy and training data is scarce. In this paper, we propose a novel approach (Neural Body Fitting (NBF)). It integrates a statistical body model within a CNN, leveraging reliable bottom-up semantic body part segmentation and robust top-down body model constraints. NBF is fully differentiable and can be trained using 2D and 3D annotations. In detailed experiments, we analyze how the components of our model affect performance, especially the use of part segmentations as an explicit intermediate representation, and present a robust, efficiently trainable framework for 3D human pose estimation from 2D images with competitive results on standard benchmarks. Code is available at https://github.com/mohomran/neural_body_fitting

ps

arXiv code [BibTex]


Thumb xl screen shot 2018 09 19 at 09.33.59
ClusterNet: Instance Segmentation in RGB-D Images

Shao, L., Tian, Y., Bohg, J.

arXiv, September 2018, Submitted to ICRA'19 (article) Submitted

Abstract
We propose a method for instance-level segmentation that uses RGB-D data as input and provides detailed information about the location, geometry and number of {\em individual\/} objects in the scene. This level of understanding is fundamental for autonomous robots. It enables safe and robust decision-making under the large uncertainty of the real-world. In our model, we propose to use the first and second order moments of the object occupancy function to represent an object instance. We train an hourglass Deep Neural Network (DNN) where each pixel in the output votes for the 3D position of the corresponding object center and for the object's size and pose. The final instance segmentation is achieved through clustering in the space of moments. The object-centric training loss is defined on the output of the clustering. Our method outperforms the state-of-the-art instance segmentation method on our synthesized dataset. We show that our method generalizes well on real-world data achieving visually better segmentation results.

am

link (url) [BibTex]

link (url) [BibTex]


Thumb xl grasping
Leveraging Contact Forces for Learning to Grasp

Merzic, H., Bogdanovic, M., Kappler, D., Righetti, L., Bohg, J.

arXiv, September 2018, Submitted to ICRA'19 (article) Submitted

Abstract
Grasping objects under uncertainty remains an open problem in robotics research. This uncertainty is often due to noisy or partial observations of the object pose or shape. To enable a robot to react appropriately to unforeseen effects, it is crucial that it continuously takes sensor feedback into account. While visual feedback is important for inferring a grasp pose and reaching for an object, contact feedback offers valuable information during manipulation and grasp acquisition. In this paper, we use model-free deep reinforcement learning to synthesize control policies that exploit contact sensing to generate robust grasping under uncertainty. We demonstrate our approach on a multi-fingered hand that exhibits more complex finger coordination than the commonly used two- fingered grippers. We conduct extensive experiments in order to assess the performance of the learned policies, with and without contact sensing. While it is possible to learn grasping policies without contact sensing, our results suggest that contact feedback allows for a significant improvement of grasping robustness under object pose uncertainty and for objects with a complex shape.

am

video arXiv [BibTex]


Thumb xl eccv pascal results  thumbnail
Deep Directional Statistics: Pose Estimation with Uncertainty Quantification

Prokudin, S., Gehler, P., Nowozin, S.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
Modern deep learning systems successfully solve many perception tasks such as object pose estimation when the input image is of high quality. However, in challenging imaging conditions such as on low resolution images or when the image is corrupted by imaging artifacts, current systems degrade considerably in accuracy. While a loss in performance is unavoidable we would like our models to quantify their uncertainty in order to achieve robustness against images of varying quality. Probabilistic deep learning models combine the expressive power of deep learning with uncertainty quantification. In this paper, we propose a novel probabilistic deep learning model for the task of angular regression. Our model uses von Mises distributions to predict a distribution over object pose angle. Whereas a single von Mises distribution is making strong assumptions about the shape of the distribution, we extend the basic model to predict a mixture of von Mises distributions. We show how to learn a mixture model using a finite and infinite number of mixture components. Our model allow for likelihood-based training and efficient inference at test time. We demonstrate on a number of challenging pose estimation datasets that our model produces calibrated probability predictions and competitive or superior point estimates compared to the current state-of-the-art.

ps

code pdf [BibTex]

code pdf [BibTex]


Thumb xl beneccv18
SphereNet: Learning Spherical Representations for Detection and Classification in Omnidirectional Images

Coors, B., Condurache, A. P., Geiger, A.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
Omnidirectional cameras offer great benefits over classical cameras wherever a wide field of view is essential, such as in virtual reality applications or in autonomous robots. Unfortunately, standard convolutional neural networks are not well suited for this scenario as the natural projection surface is a sphere which cannot be unwrapped to a plane without introducing significant distortions, particularly in the polar regions. In this work, we present SphereNet, a novel deep learning framework which encodes invariance against such distortions explicitly into convolutional neural networks. Towards this goal, SphereNet adapts the sampling locations of the convolutional filters, effectively reversing distortions, and wraps the filters around the sphere. By building on regular convolutions, SphereNet enables the transfer of existing perspective convolutional neural network models to the omnidirectional case. We demonstrate the effectiveness of our method on the tasks of image classification and object detection, exploiting two newly created semi-synthetic and real-world omnidirectional datasets.

avg

pdf suppmat Project Page [BibTex]


Thumb xl vip
Recovering Accurate 3D Human Pose in The Wild Using IMUs and a Moving Camera

Marcard, T. V., Henschel, R., Black, M. J., Rosenhahn, B., Pons-Moll, G.

European Conference on Computer Vision (ECCV), September 2018 (conference)

Abstract
In this work, we propose a method that combines a single hand-held camera and a set of Inertial Measurement Units (IMUs) attached at the body limbs to estimate accurate 3D poses in the wild. This poses many new challenges: the moving camera, heading drift, cluttered background, occlusions and many people visible in the video. We associate 2D pose detections in each image to the corresponding IMU-equipped persons by solving a novel graph based optimization problem that forces 3D to 2D coherency within a frame and across long range frames. Given associations, we jointly optimize the pose of a statistical body model, the camera pose and heading drift using a continuous optimization framework. We validated our method on the TotalCapture dataset, which provides video and IMU synchronized with ground truth. We obtain an accuracy of 26mm, which makes it accurate enough to serve as a benchmark for image-based 3D pose estimation in the wild. Using our method, we recorded 3D Poses in the Wild (3DPW ), a new dataset consisting of more than 51; 000 frames with accurate 3D pose in challenging sequences, including walking in the city, going up-stairs, having co ffee or taking the bus. We make the reconstructed 3D poses, video, IMU and 3D models available for research purposes at http://virtualhumans.mpi-inf.mpg.de/3DPW.

ps

pdf SupMat data project [BibTex]

pdf SupMat data project [BibTex]


Thumb xl sample3 merge black
Learning an Infant Body Model from RGB-D Data for Accurate Full Body Motion Analysis

Hesse, N., Pujades, S., Romero, J., Black, M. J., Bodensteiner, C., Arens, M., Hofmann, U. G., Tacke, U., Hadders-Algra, M., Weinberger, R., Muller-Felber, W., Schroeder, A. S.

In Int. Conf. on Medical Image Computing and Computer Assisted Intervention (MICCAI), September 2018 (inproceedings)

Abstract
Infant motion analysis enables early detection of neurodevelopmental disorders like cerebral palsy (CP). Diagnosis, however, is challenging, requiring expert human judgement. An automated solution would be beneficial but requires the accurate capture of 3D full-body movements. To that end, we develop a non-intrusive, low-cost, lightweight acquisition system that captures the shape and motion of infants. Going beyond work on modeling adult body shape, we learn a 3D Skinned Multi-Infant Linear body model (SMIL) from noisy, low-quality, and incomplete RGB-D data. We demonstrate the capture of shape and motion with 37 infants in a clinical environment. Quantitative experiments show that SMIL faithfully represents the data and properly factorizes the shape and pose of the infants. With a case study based on general movement assessment (GMA), we demonstrate that SMIL captures enough information to allow medical assessment. SMIL provides a new tool and a step towards a fully automatic system for GMA.

ps

pdf Project page video extended arXiv version [BibTex]

pdf Project page video extended arXiv version [BibTex]


Thumb xl toc image
Diffusion Measurements of Swimming Enzymes with Fluorescence Correlation Spectroscopy

Günther, J., Börsch, M., Fischer, P.

Accounts of Chemical Research, August 2018, PMID: 30160941 (article)

Abstract
ConspectusSelf-propelled chemical motors are chemically powered micro- or nanosized swimmers. The energy required for these motors’ active motion derives from catalytic chemical reactions and the transformation of a fuel dissolved in the solution. While self-propulsion is now well established for larger particles, it is still unclear if enzymes, nature’s nanometer-sized catalysts, are potentially also self-powered nanomotors. Because of its small size, any increase in an enzyme’s diffusion due to active self-propulsion must be observed on top of the enzyme’s passive Brownian motion, which dominates at this scale. Fluorescence correlation spectroscopy (FCS) is a sensitive method to quantify the diffusion properties of single fluorescently labeled molecules in solution. FCS experiments have shown a general increase in the diffusion constant of a number of enzymes when the enzyme is catalytically active. Diffusion enhancements after addition of the enzyme’s substrate (and sometimes its inhibitor) of up to 80\% have been reported, which is at least 1 order of magnitude higher than what theory would predict. However, many factors contribute to the FCS signal and in particular the shape of the autocorrelation function, which underlies diffusion measurements by fluorescence correlation spectroscopy. These effects need to be considered to establish if and by how much the catalytic activity changes an enzyme’s diffusion.We carefully review phenomena that can play a role in FCS experiments and the determination of enzyme diffusion, including the dissociation of enzyme oligomers upon interaction with the substrate, surface binding of the enzyme to glass during the experiment, conformational changes upon binding, and quenching of the fluorophore. We show that these effects can cause changes in the FCS signal that behave similar to an increase in diffusion. However, in the case of the enzymes F1-ATPase and alkaline phosphatase, we demonstrate that there is no measurable increase in enzyme diffusion. Rather, dissociation and conformational changes account for the changes in the FCS signal in the former and fluorophore quenching in the latter. Within the experimental accuracy of our FCS measurements, we do not observe any change in diffusion due to activity for the enzymes we have investigated.We suggest useful control experiments and additional tests for future FCS experiments that should help establish if the observed diffusion enhancement is real or if it is due to an experimental or data analysis artifact. We show that fluorescence lifetime and mean intensity measurements are essential in order to identify the nature of the observed changes in the autocorrelation function. While it is clear from theory that chemically active enzymes should also act as self-propelled nanomotors, our FCS measurements show that the associated increase in diffusion is much smaller than previously reported. Further experiments are needed to quantify the contribution of the enzymes’ catalytic activity to their self-propulsion. We hope that our findings help to establish a useful protocol for future FCS studies in this field and help establish by how much the diffusion of an enzyme is enhanced through catalytic activity.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


Thumb xl toc imagen
Uphill production of dihydrogen by enzymatic oxidation of glucose without an external energy source

Suraniti, E., Merzeau, P., Roche, J., Gounel, S., Mark, A. G., Fischer, P., Mano, N., Kuhn, A.

Nature Communications, 9(1):3229, August 2018 (article)

Abstract
Chemical systems do not allow the coupling of energy from several simple reactions to drive a subsequent reaction, which takes place in the same medium and leads to a product with a higher energy than the one released during the first reaction. Gibbs energy considerations thus are not favorable to drive e.g., water splitting by the direct oxidation of glucose as a model reaction. Here, we show that it is nevertheless possible to carry out such an energetically uphill reaction, if the electrons released in the oxidation reaction are temporarily stored in an electromagnetic system, which is then used to raise the electrons' potential energy so that they can power the electrolysis of water in a second step. We thereby demonstrate the general concept that lower energy delivering chemical reactions can be used to enable the formation of higher energy consuming reaction products in a closed system.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


Thumb xl toc image
Chemical micromotors self-assemble and self-propel by spontaneous symmetry breaking

Yu, T., Chuphal, P., Thakur, S., Reigh, S. Y., Singh, D. P., Fischer, P.

Chem. Commun., pages: -, The Royal Society of Chemistry, August 2018 (article)

Abstract
Self-propelling chemical motors have thus far required the fabrication of Janus particles with an asymmetric catalyst distribution. Here{,} we demonstrate that simple{,} isotropic colloids can spontaneously assemble to yield dimer motors that self-propel. In a mixture of isotropic titanium dioxide colloids with photo-chemical catalytic activity and passive silica colloids{,} light illumination causes diffusiophoretic attractions between the active and passive particles and leads to the formation of dimers. The dimers constitute a symmetry-broken motor{,} whose dynamics can be fully controlled by the illumination conditions. Computer simulations reproduce the dynamics of the colloids and are in good agreement with experiments. The current work presents a simple route to obtain large numbers of self-propelling chemical motors from a dispersion of spherically symmetric colloids through spontaneous symmetry breaking.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Statistical Modelling of Fingertip Deformations and Contact Forces during Tactile Interaction

Gueorguiev, D., Tzionas, D., Pacchierotti, C., Black, M. J., Kuchenbecker, K. J.

Extended abstract presented at the 2018 CSF Hand, Brain and Technology conference (HBT), August 2018 (misc)

Abstract
Little is known about the shape and properties of the human finger during haptic interaction, even though these are essential parameters for controlling wearable finger devices and deliver realistic tactile feedback. This study explores a framework for four-dimensional scanning (3D over time) and modelling of finger-surface interactions, aiming to capture the motion and deformations of the entire finger with high resolution while simultaneously recording the interfacial forces at the contact. Preliminary results show that when the fingertip is actively pressing a rigid surface, it undergoes lateral expansion and proximal/distal bending, deformations that cannot be captured by imaging of the contact area alone. Therefore, we are currently capturing a dataset that will enable us to create a statistical model of the finger’s deformations and predict the contact forces induced by tactile interaction with objects. This technique could improve current methods for tactile rendering in wearable haptic devices, which rely on general physical modelling of the skin’s compliance, by developing an accurate model of the variations in finger properties across the human population. The availability of such a model will also enable a more realistic simulation of virtual finger behaviour in virtual reality (VR) environments, as well as the ability to accurately model a specific user’s finger from lower resolution data. It may also be relevant for inferring the physical properties of the underlying tissue from observing the surface mesh deformations, as previously shown for body tissues.

hi

Project Page [BibTex]

Project Page [BibTex]


no image
Learning-Based Robust Model Predictive Control with State-Dependent Uncertainty

Soloperto, R., Müller, M. A., Trimpe, S., Allgöwer, F.

In IFAC Conference on Nonlinear Model Predictive Control, Madison, Wisconsin, USA, 6th IFAC Conference on Nonlinear Model Predictive Control, August 2018 (inproceedings) Accepted

ics

[BibTex]

[BibTex]


Thumb xl aircap ca 3
Decentralized MPC based Obstacle Avoidance for Multi-Robot Target Tracking Scenarios

Tallamraju, R., Rajappa, S., Black, M. J., Karlapalem, K., Ahmad, A.

2018 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR), pages: 1-8, IEEE, August 2018 (conference)

Abstract
In this work, we consider the problem of decentralized multi-robot target tracking and obstacle avoidance in dynamic environments. Each robot executes a local motion planning algorithm which is based on model predictive control (MPC). The planner is designed as a quadratic program, subject to constraints on robot dynamics and obstacle avoidance. Repulsive potential field functions are employed to avoid obstacles. The novelty of our approach lies in embedding these non-linear potential field functions as constraints within a convex optimization framework. Our method convexifies nonconvex constraints and dependencies, by replacing them as pre-computed external input forces in robot dynamics. The proposed algorithm additionally incorporates different methods to avoid field local minima problems associated with using potential field functions in planning. The motion planner does not enforce predefined trajectories or any formation geometry on the robots and is a comprehensive solution for cooperative obstacle avoidance in the context of multi-robot target tracking. We perform simulation studies for different scenarios to showcase the convergence and efficacy of the proposed algorithm.

ps

Published Version link (url) DOI Project Page [BibTex]

Published Version link (url) DOI Project Page [BibTex]


Thumb xl toc image
Gait learning for soft microrobots controlled by light fields

Rohr, A. V., Trimpe, S., Marco, A., Fischer, P., Palagi, S.

In Proceeding of the International Conference on Intelligent Robots and Systems, International Conference on Intelligent Robots and Systems (IROS), July 2018 (inproceedings) To be published

Abstract
Soft microrobots based on photoresponsive materials and controlled by light fields can generate a variety of different gaits. This inherent flexibility can be exploited to maximize their locomotion performance in a given environment and used to adapt them to changing environments. However, because of the lack of accurate locomotion models, and given the intrinsic variability among microrobots, analytical control design is not possible. Common data-driven approaches, on the other hand, require running prohibitive numbers of experiments and lead to very sample-specific results. Here we propose a probabilistic learning approach for light-controlled soft microrobots based on Bayesian Optimization (BO) and Gaussian Processes (GPs). The proposed approach results in a learning scheme that is highly data-efficient, enabling gait optimization with a limited experimental budget, and robust against differences among microrobot samples. These features are obtained by designing the learning scheme through the comparison of different GP priors and BO settings on a semisynthetic data set. The developed learning scheme is validated in microrobot experiments, resulting in a 115% improvement in a microrobot’s locomotion performance with an experimental budget of only 20 tests. These encouraging results lead the way toward self-adaptive microrobotic systems based on lightcontrolled soft microrobots and probabilistic learning control.

ics pf

arXiv Project Page [BibTex]

arXiv Project Page [BibTex]


Thumb xl toc image
A machine from machines

Fischer, P.

Nature Physics, July 2018 (article)

Abstract
Building spinning microrotors that self-assemble and synchronize to form a gear sounds like an impossible feat. However, it has now been achieved using only a single type of building block -- a colloid that self-propels.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


Thumb xl toc image
Chemotaxis of Active Janus Nanoparticles

Popescu, M. N., Uspal, W. E., Bechinger, C., Fischer, P.

Nano Letters, July 2018, PMID: 30047271 (article)

Abstract
While colloids and molecules in solution exhibit passive Brownian motion, particles that are partially covered with a catalyst, which promotes the transformation of a fuel dissolved in the solution, can actively move. These active Janus particles are known as “chemical nanomotors” or self-propelling “swimmers” and have been realized with a range of catalysts, sizes, and particle geometries. Because their active translation depends on the fuel concentration, one expects that active colloidal particles should also be able to swim toward a fuel source. Synthesizing and engineering nanoparticles with distinct chemotactic properties may enable important developments, such as particles that can autonomously swim along a pH gradient toward a tumor. Chemotaxis requires that the particles possess an active coupling of their orientation to a chemical gradient. In this Perspective we provide a simple, intuitive description of the underlying mechanisms for chemotaxis, as well as the means to analyze and classify active particles that can show positive or negative chemotaxis. The classification provides guidance for engineering a specific response and is a useful organizing framework for the quantitative analysis and modeling of chemotactic behaviors. Chemotaxis is emerging as an important focus area in the field of active colloids and promises a number of fascinating applications for nanoparticles and particle-based delivery.

pf

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Kernel Recursive ABC: Point Estimation with Intractable Likelihood

Kajihara, T., Kanagawa, M., Yamazaki, K., Fukumizu, K.

Proceedings of the 35th International Conference on Machine Learning, pages: 2405-2414, PMLR, July 2018 (proceedings)

Abstract
We propose a novel approach to parameter estimation for simulator-based statistical models with intractable likelihood. Our proposed method involves recursive application of kernel ABC and kernel herding to the same observed data. We provide a theoretical explanation regarding why the approach works, showing (for the population setting) that, under a certain assumption, point estimates obtained with this method converge to the true parameter, as recursion proceeds. We have conducted a variety of numerical experiments, including parameter estimation for a real-world pedestrian flow simulator, and show that in most cases our method outperforms existing approaches.

pn

Paper [BibTex]

Paper [BibTex]


Thumb xl emthy 01
3D-Printed Biodegradable Microswimmer for Drug Delivery and Targeted Cell Labeling

Hakan Ceylan, , I. Ceren Yasa, , Oncay Yasa, , Ahmet Fatih Tabak, , Joshua Giltinan, , Sitti, M.

bioRxiv, pages: 379024, July 2018 (article)

Abstract
Miniaturization of interventional medical devices can leverage minimally invasive technologies by enabling operational resolution at cellular length scales with high precision and repeatability. Untethered micron-scale mobile robots can realize this by navigating and performing in hard-to-reach, confined and delicate inner body sites. However, such a complex task requires an integrated design and engineering strategy, where powering, control, environmental sensing, medical functionality and biodegradability need to be considered altogether. The present study reports a hydrogel-based, biodegradable microrobotic swimmer, which is responsive to the changes in its microenvironment for theranostic cargo delivery and release tasks. We design a double-helical magnetic microswimmer of 20 micrometers length, which is 3D-printed with complex geometrical and compositional features. At normal physiological concentrations, matrix metalloproteinase-2 (MMP-2) enzyme can entirely degrade the microswimmer body in 118 h to solubilized non-toxic products. The microswimmer can respond to the pathological concentrations of MMP-2 by swelling and thereby accelerating the release kinetics of the drug payload. Anti-ErbB 2 antibody-tagged magnetic nanoparticles released from the degraded microswimmers serve for targeted labeling of SKBR3 breast cancer cells to realize the potential of medical imaging of local tissue sites following the therapeutic intervention. These results represent a leap forward toward clinical medical microrobots that are capable of sensing, responding to the local pathological information, and performing specific therapeutic and diagnostic tasks as orderly executed operations using their smart composite material architectures.

pi

DOI [BibTex]


Thumb xl teaser image
Programmable collective behavior in dynamically self-assembled mobile microrobotic swarms

B Yigit, , Y Alapan, , Sitti, M.

arXiv, July 2018 (article)

Abstract
Collective control of mobile microrobotic swarms is indispensable for their potential high-impact applications in targeted drug delivery, medical diagnostics, parallel micromanipulation, and environmental sensing and remediation. Lack of on-board computational and sensing capabilities in current microrobotic systems necessitates use of physical interactions among individual microrobots for local physical communication and cooperation. Here, we show that mobile microrobotic swarms with well-defined collective behavior can be designed by engineering magnetic interactions among individual units. Microrobots, consisting of a linear chain of self-assembled magnetic microparticles, locomote on surfaces in response to a precessing magnetic field. Control over the direction of precessing magnetic field allows engineering attractive and repulsive interactions among microrobots and, thus, collective order with well-defined spatial organization and parallel operation over macroscale distances (~ 1 cm). These microrobotic swarms can be guided through confined spaces, while preserving microrobot morphology and function. These swarms can further achieve directional transport of large cargoes on surfaces and small cargoes in bulk fluids. Described design approach, exploiting physical interactions among individual robots, enables facile and rapid formation of self-organized and reconfigurable microrobotic swarms with programmable collective order.

pi

link (url) [BibTex]


no image
Learning-based solution to phase error correction in T2*-weighted GRE scans

Loktyushin, A., Ehses, P., Schölkopf, B., Scheffler, K.

1st International conference on Medical Imaging with Deep Learning (MIDL), July 2018 (conference)

ei

link (url) [BibTex]

link (url) [BibTex]


no image
Intrinsic disentanglement: an invariance view for deep generative models

Besserve, M., Sun, R., Schölkopf, B.

Workshop on Theoretical Foundations and Applications of Deep Generative Models at ICML, July 2018 (conference)

ei

PDF [BibTex]

PDF [BibTex]