Abstract-Scene understanding is a necessary prerequisite for robots acting autonomously in complex environments. Low-cost RGB-D cameras such as Microsoft Kinect enabled new methods for analyzing indoor scenes and are now ubiquitously used in indoor robotics. We investigate strategies for efficient pixelwise object class labeling of indoor scenes that combine both pretrained semantic features transferred from a large color image dataset and geometric features, computed relative to the room structures, including a novel distance-from-wall feature, which encodes the proximity of scene points to a detected major wall of the room. We evaluate our approach on the popular NYU v2 dataset. Several deep learning models are tested, which are designed to exploit different characteristics of the data. This includes feature learning with two different pooling sizes. Our results indicate that combining semantic and geometric features yields significantly improved results for the task of object class segmentation.

Abstract-We propose a new approach for the segmentation of 3-D point clouds into geometric surfaces using adaptive surface models. Starting from an initial configuration, the algorithm converges to a stable segmentation through a new iterative splitand-merge procedure, which includes an adaptive mechanism for the creation and removal of segments. This allows the segmentation to adjust to changing input data along the movie, leading to stable, temporally coherent, and traceable segments. We tested the method on a large variety of data acquired with different range imaging devices, including a structured-light sensor and a time-of-flight camera, and successfully segmented the videos into surface segments. We further demonstrated the feasibility of the approach using quantitative evaluations based on ground-truth data.

Abstract-In this paper we present an automated system that is able to track and grasp a moving object within the workspace of a manipulator using range images acquired with a Microsoft Kinect sensor. Realtime tracking is achieved by a geometric particle filter on the affine group. Based on the tracked output, the pose of a 7-DoF WAM robotic arm is continuously updated using dynamic motor primitives until a distance measure between the tracked object and the gripper mounted on the arm is below a threshold. Then, it closes its three fingers and grasps the object. The tracker works in realtime and is robust to noise and partial occlusions. Using only the depth data makes our tracker independent of texture which is one of the key design goals in our approach. An experimental evaluation is provided along with a comparison of the proposed tracker with state-of-the-art approaches, including the OpenNItracker. The developed system is integrated with ROS and made available as part of IRI's ROS stack.

Abstract-Hand-crafted feature functions are usually designed based on the domain knowledge of a presumably controlled environment and often fail to generalize, as the statistics of realworld data cannot always be modeled correctly. Data-driven feature learning methods, on the other hand, have emerged as an alternative that often generalize better in uncontrolled environments. We present a simple, yet robust, 2D convolutional neural network extended to a concatenated 3D network that learns to extract features from the spatio-temporal domain of raw video data. The resulting network model is used for contentbased recognition of videos. Relying on a 2D convolutional neural network allows us to exploit a pretrained network as a descriptor that yielded the best results on the largest and challenging ILSVRC-2014 dataset. Experimental results on commonly used benchmarking video datasets demonstrate that our results are state-of-the-art in terms of accuracy and computational time without requiring any preprocessing (e.g., optic flow) or a priori knowledge on data capture (e.g., camera motion estimation), which makes it more general and flexible than other approaches. Our implementation is made available.

Deep learning is a type of machine perception method that attempts to model highlevel abstractions in data and encode them into a compact and robust representation. Such representations have found immense usage in applications related to computer vision. In this chapter we introduce two such applications, i.e., semantic segmentation of images and action recognition in videos. These applications are of fundamental importance for human-centered environment perception.

Abstract-Reliable object discovery in realistic indoor scenes is a necessity for many computer vision and service robot applications. In these scenes, semantic segmentation methods have made huge advances in recent years. Such methods can provide useful prior information for object discovery by removing false positives and by delineating object boundaries. We propose a novel method that combines bottom-up object discovery and semantic priors for producing generic object candidates in RGB-D images. We use a deep learning method for semantic segmentation to classify colour and depth superpixels into meaningful categories. Separately for each category, we use saliency to estimate the location and scale of objects, and superpixels to find their precise boundaries. Finally, object candidates of all categories are combined and ranked. We evaluate our approach on the NYU Depth V2 dataset and show that we outperform other state-of-the-art object discovery methods in terms of recall.

Abstract-Detecting objects in cluttered scenes is a necessary step for many robotic tasks and facilitates the interaction of the robot with its environment. Because of the availability of efficient 3D sensing devices as the Kinect, methods for the recognition of objects in 3D point clouds have gained importance during the last years. In this paper, we propose a new supervised learning approach for the recognition of objects from 3D point clouds using Conditional Random Fields, a type of discriminative, undirected probabilistic graphical model. The various features and contextual relations of the objects are described by the potential functions in the graph. Our method allows for learning and inference from unorganized point clouds of arbitrary sizes and shows significant benefit in terms of computational speed during prediction when compared to a state-of-the-art approach based on constrained optimization.

Abstract:A novel framework for joint segmentation and tracking in depth videos of object surfaces is presented. Initially, the 3D colored point cloud obtained using the Kinect camera is used to segment the scene into surface patches, defined by quadratic functions. The computed segments together with their functional descriptions are then used to partition the depth image of the subsequent frame in a consistent manner with respect to the precedent frame. This way, solutions established in previous frames can be reused which improves the efficiency of the algorithm and the coherency of the segmentations along the movie. The algorithm is tested for scenes showing human and robot manipulations of objects. We demonstrate that the method can successfully segment and track the human/robot arm and object surfaces along the manipulations. The performance is evaluated quantitatively by measuring the temporal coherency of the segmentations and the segmentation covering using ground truth. The method provides a visual front-end designed for robotic applications, and can potentially be used in the context of manipulation recognition, visual servoing, and robot-grasping tasks.