Towards a real-time 3D object recognition pipeline on mobile GPGPU computing platforms using low-cost RGB-D sensors

In this project, we propose the implementation of a 3D object recognition system which will be optimized to operate under demanding time constraints. The system must be robust so that objects can be recognized properly in poor light conditions and cluttered scenes with significant levels of occlusion. An important requirement must be met: the system must exhibit a reasonable performance running on a low power consumption mobile GPU computing platform (NVIDIA Jetson TK1) so that it can be integrated in mobile robotics systems, ambient intelligence or ambient assisted living applications. The acquisition system is based on the use of color and depth (RGB-D) data streams provided by low-cost 3D sensors like Microsoft Kinect or PrimeSense Carmine. The range of algorithms and applications to be implemented and integrated will be quite broad, ranging from the acquisition, outlier removal or filtering of the input data and the segmentation or characterization of regions of interest in the scene to the very object recognition and pose estimation. Furthermore, in order to validate the proposed system, we will create a 3D object dataset. It will be composed by a set of 3D models, reconstructed from common household objects, as well as a handful of test scenes in which those objects appear. The scenes will be characterized by different levels of occlusion, diverse distances from the elements to the sensor and variations on the pose of the target objects. The creation of this dataset implies the additional development of 3D data acquisition and 3D object reconstruction applications. The resulting system has many possible applications, ranging from mobile robot navigation and semantic scene labeling to human-computer interaction (HCI) systems based on visual information.

3DCOMET: 3D compression methods test dataset

The use of 3D data in mobile robotics applications provides valuable information about the robot’s environment. However usually the huge amount of 3D information is difficult to manage due to the fact that the robot storage system and computing capabilities are insufficient. Therefore, a data compression method is necessary to store and process this information while preserving as much information as possible. A few methods have been proposed to compress 3D information. Nevertheless, there does not exist a consistent public benchmark for comparing the results (compression level, distance reconstructed error, etc.) obtained with different methods. In this paper, we propose a dataset composed of a set of 3D point clouds with different structure and texture variability to evaluate the results obtained from 3D data compression methods. We also provide useful tools for comparing compression methods, using as a baseline the results obtained by existing relevant compression methods.

Evaluation of sampling method effects in 3D non-rigid registration

Since the beginning of 3D computer vision problems, the use of techniques to reduce the data to make it treatable preserving the important aspects of the scene has been necessary. Currently, with the new low-cost RGB-D sensors, which provide a stream of color and 3D data of approximately 30 frames per second, this is getting more relevance. Many applications make use of these sensors and need a preprocessing to downsample the data in order to either reduce the processing time or improve the data (e.g., reducing noise or enhancing the important features). In this paper, we present a comparison of different downsampling techniques which are based on different principles. Concretely, five different downsampling methods are included: a bilinear-based method, a normal-based, a color-based, a combination of the normal and color-based samplings, and a growing neural gas (GNG)-based approach. For the comparison, two different models have been used acquired with the Blensor software. Moreover, to evaluate the effect of the downsampling in a real application, a 3D non-rigid registration is performed with the data sampled. From the experimentation we can conclude that depending on the purpose of the application some kernels of the sampling methods can improve drastically the results. Bilinear- and GNG-based methods provide homogeneous point clouds, but color-based and normal-based provide datasets with higher density of points in areas with specific features. In the non-rigid application, if a color-based sampled point cloud is used, it is possible to properly register two datasets for cases where intensity data are relevant in the model and outperform the results if only a homogeneous sampling is used.

Large scale SLAM with visual features

Comunicación presentada en el X Workshop of Physical Agents, Cáceres, 10-11 septiembre 2009.

JavaVis: An Integrated Computer Vision Library for Teaching Computer Vision

In this article, we present a new framework oriented to teach Computer Vision related subjects called JavaVis. It is a computer vision library divided in three main areas: 2D package is featured for classical computer vision processing; 3D package, which includes a complete 3D geometric toolset, is used for 3D vision computing; Desktop package comprises a tool for graphic designing and testing of new algorithms. JavaVis is designed to be easy to use, both for launching and testing existing algorithms and for developing new ones.

μ-MAR: Multiplane 3D Marker based Registration for depth-sensing cameras

Many applications including object reconstruction, robot guidance, and. scene mapping require the registration of multiple views from a scene to generate a complete geometric and appearance model of it. In real situations, transformations between views are unknown and it is necessary to apply expert inference to estimate them. In the last few years, the emergence of low-cost depth-sensing cameras has strengthened the research on this topic, motivating a plethora of new applications. Although they have enough resolution and accuracy for many applications, some situations may not be solved with general state-of-the-art registration methods due to the signal-to-noise ratio (SNR) and the resolution of the data provided. The problem of working with low SNR data, in general terms, may appear in any 3D system, then it is necessary to propose novel solutions in this aspect. In this paper, we propose a method, μ-MAR, able to both coarse and fine register sets of 3D points provided by low-cost depth-sensing cameras, despite it is not restricted to these sensors, into a common coordinate system. The method is able to overcome the noisy data problem by means of using a model-based solution of multiplane registration. Specifically, it iteratively registers 3D markers composed by multiple planes extracted from points of multiple views of the scene. As the markers and the object of interest are static in the scenario, the transformations obtained for the markers are applied to the object in order to reconstruct it. Experiments have been performed using synthetic and real data. The synthetic data allows a qualitative and quantitative evaluation by means of visual inspection and Hausdorff distance respectively. The real data experiments show the performance of the proposal using data acquired by a Primesense Carmine RGB-D sensor. The method has been compared to several state-of-the-art methods. The results show the good performance of the μ-MAR to register objects with high accuracy in presence of noisy data outperforming the existing methods.

3D Visual Data-Driven Spatiotemporal Deformations for Non-Rigid Object Grasping Using Robot Hands

Sensing techniques are important for solving problems of uncertainty inherent to intelligent grasping tasks. The main goal here is to present a visual sensing system based on range imaging technology for robot manipulation of non-rigid objects. Our proposal provides a suitable visual perception system of complex grasping tasks to support a robot controller when other sensor systems, such as tactile and force, are not able to obtain useful data relevant to the grasping manipulation task. In particular, a new visual approach based on RGBD data was implemented to help a robot controller carry out intelligent manipulation tasks with flexible objects. The proposed method supervises the interaction between the grasped object and the robot hand in order to avoid poor contact between the fingertips and an object when there is neither force nor pressure data. This new approach is also used to measure changes to the shape of an object’s surfaces and so allows us to find deformations caused by inappropriate pressure being applied by the hand’s fingers. Test was carried out for grasping tasks involving several flexible household objects with a multi-fingered robot hand working in real time. Our approach generates pulses from the deformation detection method and sends an event message to the robot controller when surface deformation is detected. In comparison with other methods, the obtained results reveal that our visual pipeline does not use deformations models of objects and materials, as well as the approach works well both planar and 3D household objects in real time. In addition, our method does not depend on the pose of the robot hand because the location of the reference system is computed from a recognition process of a pattern located place at the robot forearm. The presented experiments demonstrate that the proposed method accomplishes a good monitoring of grasping task with several objects and different grasping configurations in indoor environments.

GPGPU implementation of growing neural gas: application to 3D scene reconstruction

Self-organising neural models have the ability to provide a good representation of the input space. In particular the Growing Neural Gas (GNG) is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation. However, this type of learning is time-consuming, especially for high-dimensional input data. Since real applications often work under time constraints, it is necessary to adapt the learning process in order to complete it in a predefined time. This paper proposes a Graphics Processing Unit (GPU) parallel implementation of the GNG with Compute Unified Device Architecture (CUDA). In contrast to existing algorithms, the proposed GPU implementation allows the acceleration of the learning process keeping a good quality of representation. Comparative experiments using iterative, parallel and hybrid implementations are carried out to demonstrate the effectiveness of CUDA implementation. The results show that GNG learning with the proposed implementation achieves a speed-up of 6× compared with the single-threaded CPU implementation. GPU implementation has also been applied to a real application with time constraints: acceleration of 3D scene reconstruction for egomotion, in order to validate the proposal.

Real-time 3D semi-local surface patch extraction using GPGPU

Feature vectors can be anything from simple surface normals to more complex feature descriptors. Feature extraction is important to solve various computer vision problems: e.g. registration, object recognition and scene understanding. Most of these techniques cannot be computed online due to their complexity and the context where they are applied. Therefore, computing these features in real-time for many points in the scene is impossible. In this work, a hardware-based implementation of 3D feature extraction and 3D object recognition is proposed to accelerate these methods and therefore the entire pipeline of RGBD based computer vision systems where such features are typically used. The use of a GPU as a general purpose processor can achieve considerable speed-ups compared with a CPU implementation. In this work, advantageous results are obtained using the GPU to accelerate the computation of a 3D descriptor based on the calculation of 3D semi-local surface patches of partial views. This allows descriptor computation at several points of a scene in real-time. Benefits of the accelerated descriptor have been demonstrated in object recognition tasks. Source code will be made publicly available as contribution to the Open Source Point Cloud Library.

3D Maps Representation Using GNG

Current RGB-D sensors provide a big amount of valuable information for mobile robotics tasks like 3D map reconstruction, but the storage and processing of the incremental data provided by the different sensors through time quickly become unmanageable. In this work, we focus on 3D maps representation and propose the use of the Growing Neural Gas (GNG) network as a model to represent 3D input data. GNG method is able to represent the input data with a desired amount of neurons or resolution while preserving the topology of the input space. Experiments show how GNG method yields a better input space adaptation than other state-of-the-art 3D map representation methods.

Discontinuity spacing analysis in rock masses using 3D point clouds

The complete characterization of rock masses implies the acquisition of information of both, the materials which compose the rock mass and the discontinuities which divide the outcrop. Recent advances in the use of remote sensing techniques – such as Light Detection and Ranging (LiDAR) – allow the accurate and dense acquisition of 3D information that can be used for the characterization of discontinuities. This work presents a novel methodology which allows the calculation of the normal spacing of persistent and non-persistent discontinuity sets using 3D point cloud datasets considering the three dimensional relationships between clusters. This approach requires that the 3D dataset has been previously classified. This implies that discontinuity sets are previously extracted, every single point is labeled with its corresponding discontinuity set and every exposed planar surface is analytically calculated. Then, for each discontinuity set the method calculates the normal spacing between an exposed plane and its nearest one considering 3D space relationship. This link between planes is obtained calculating for every point its nearest point member of the same discontinuity set, which provides its nearest plane. This allows calculating the normal spacing for every plane. Finally, the normal spacing is calculated as the mean value of all the normal spacings for each discontinuity set. The methodology is validated through three cases of study using synthetic data and 3D laser scanning datasets. The first case illustrates the fundamentals and the performance of the proposed methodology. The second and the third cases of study correspond to two rock slopes for which datasets were acquired using a 3D laser scanner. The second case study has shown that results obtained from the traditional and the proposed approaches are reasonably similar. Nevertheless, a discrepancy between both approaches has been found when the exposed planes members of a discontinuity set were hard to identify and when the planes pairing was difficult to establish during the fieldwork campaign. The third case study also has evidenced that when the number of identified exposed planes is high, the calculated normal spacing using the proposed approach is minor than those using the traditional approach.

Methodology based on registration techniques for representing subjects and their deformations acquired from general purpose 3D sensors

In this thesis a methodology for representing 3D subjects and their deformations in adverse situations is studied. The study is focused in providing methods based on registration techniques to improve the data in situations where the sensor is working in the limit of its sensitivity. In order to do this, it is proposed two methods to overcome the problems which can difficult the process in these conditions. First a rigid registration based on model registration is presented, where the model of 3D planar markers is used. This model is estimated using a proposed method which improves its quality by taking into account prior knowledge of the marker. To study the deformations, it is proposed a framework to combine multiple spaces in a non-rigid registration technique. This proposal improves the quality of the alignment with a more robust matching process that makes use of all available input data. Moreover, this framework allows the registration of multiple spaces simultaneously providing a more general technique. Concretely, it is instantiated using colour and location in the matching process for 3D location registration.

3D geometry of growth strata in a fault-propagation fold: insights into space–time evolution of the Crevillente Fault (Abanilla-Alicante sector), Betic Cordillera, Spain

This work presents a 3D geometric model of growth strata cropping out in a fault-propagation fold associated with the Crevillente Fault (Abanilla-Alicante sector) from the Bajo Segura Basin (eastern Betic Cordillera, southern Spain). The analysis of this 3D model enables us to unravel the along-strike and along-section variations of the growth strata, providing constraints to assess the fold development, and hence, the fault kinematic evolution in space and time. We postulate that the observed along-strike dip variations are related to lateral variation in fault displacement. Along-section variations of the progressive unconformity opening angles indicate greater fault slip in the upper Tortonian–Messinian time span; from the Messinian on, quantitative analysis of the unconformity indicate a constant or lower tectonic activity of the Crevillente Fault (Abanilla-Alicante sector); the minor abundance of striated pebbles in the Pliocene-Quaternary units could be interpreted as a decrease in the stress magnitude and consequently in the tectonic activity of the fault. At a regional scale, comparison of the growth successions cropping out in the northern and southern limits of the Bajo Segura Basin points to a southward migration of deformation in the basin. This means that the Bajo Segura Fault became active after the Crevillente Fault (Abanilla-Alicante sector), for which activity on the latter was probably decreasing according to our data. Consequently, we propose that the seismic hazard at the northern limit of the Bajo Segura Basin should be lower than at the southern limit.

Characterization of rock slopes through slope mass rating using 3D point clouds

Rock mass classification systems are widely used tools for assessing the stability of rock slopes. Their calculation requires the prior quantification of several parameters during conventional fieldwork campaigns, such as the orientation of the discontinuity sets, the main properties of the existing discontinuities and the geo-mechanical characterization of the intact rock mass, which can be time-consuming and an often risky task. Conversely, the use of relatively new remote sensing data for modelling the rock mass surface by means of 3D point clouds is changing the current investigation strategies in different rock slope engineering applications. In this paper, the main practical issues affecting the application of Slope Mass Rating (SMR) for the characterization of rock slopes from 3D point clouds are reviewed, using three case studies from an end-user point of view. To this end, the SMR adjustment factors, which were calculated from different sources of information and processes, using the different softwares, are compared with those calculated using conventional fieldwork data. In the presented analysis, special attention is paid to the differences between the SMR indexes derived from the 3D point cloud and conventional field work approaches, the main factors that determine the quality of the data and some recognized practical issues. Finally, the reliability of Slope Mass Rating for the characterization of rocky slopes is highlighted.