Search

Transparent objects are ubiquitous in daily life, making their perception and robotics manipulation important. However, they present a major challenge due to their distinct refractive and reflective properties when it comes to accurately estimating the 6D pose. To solve this, we present ReFlow6D, a novel method for transparent object 6D pose estimation that harnesses the refractive-intermediate representation. Unlike conventional approaches, our method leverages a feature space impervious to changes in RGB image space and independent of depth information. Drawing inspiration from image matting, we model the deformation of the light path through transparent objects, yielding a unique object-specific intermediate representation guided by light refraction that is independent of the environment in which objects are observed. By integrating these intermediate features into the pose estimation network, we show that ReFlow6D achieves precise 6D pose estimation of transparent objects, using only RGB images as input. Our method further introduces a novel transparent object compositing loss, fostering the generation of superior refractive-intermediate features. Empirical evaluations show that our approach significantly outperforms state-of-the-art methods on TOD and Trans32K-6D datasets. Robot grasping experiments further demonstrate that ReFlow6D's pose estimation accuracy effectively translates to real-world robotics task. The source code is available at: https://github.com/StoicGilgamesh/ReFlow6D and https://github.com/StoicGilgamesh/matting_rendering.

Continuous control tasks often involve high-dimensional, dynamic, and non-linear environments. State-of-the-art performance in these tasks is achieved through complex closed-box policies that are effective, but suffer from an inherent opacity. Interpretable policies, while generally underperforming compared to their closed-box counterparts, advantageously facilitate transparent decision-making within automated systems. Hence, their usage is often essential for diagnosing and mitigating errors, supporting ethical and legal accountability, and fostering trust among stakeholders. In this paper, we propose SMOSE, a novel method to train sparsely activated interpretable controllers, based on a top-1 Mixture-of-Experts architecture. SMOSE combines a set of interpretable decisionmakers, trained to be experts in different basic skills, and an interpretable router that assigns tasks among the experts. The training is carried out via state-of-the-art Reinforcement Learning algorithms, exploiting load-balancing techniques to ensure fair expert usage. We then distill decision trees from the weights of the router, significantly improving the ease of interpretation. We evaluate SMOSE on six benchmark environments from MuJoCo: our method outperforms recent interpretable baselines and narrows the gap with noninterpretable state-of-the-art algorithms, Comment: To be published in the Proceedings of the 39th AAAI Conference on Artificial Intelligence (AAAI-25)

Zero-shot object pose estimation enables the retrieval of object poses from images without necessitating object-specific training. In recent approaches this is facilitated by vision foundation models (VFM), which are pre-trained models that are effectively general-purpose feature extractors. The characteristics exhibited by these VFMs vary depending on the training data, network architecture, and training paradigm. The prevailing choice in this field are self-supervised Vision Transformers (ViT). This study assesses the influence of Latent Diffusion Model (LDM) backbones on zero-shot pose estimation. In order to facilitate a comparison between the two families of models on a common ground we adopt and modify a recent approach. Therefore, a template-based multi-staged method for estimating poses in a zero-shot fashion using LDMs is presented. The efficacy of the proposed approach is empirically evaluated on three standard datasets for object-specific 6DoF pose estimation. The experiments demonstrate an Average Recall improvement of up to 27% over the ViT baseline. The source code is available at: https://github.com/BvG1993/DZOP.

Oysters are a vital keystone species in coastal ecosystems, providing significant economic, environmental, and cultural benefits. As the importance of oysters grows, so does the relevance of autonomous systems for their detection and monitoring. However, current monitoring strategies often rely on destructive methods. While manual identification of oysters from video footage is non-destructive, it is time-consuming, requires expert input, and is further complicated by the challenges of the underwater environment. To address these challenges, we propose a novel pipeline using stable diffusion to augment a collected real dataset with realistic synthetic data. This method enhances the dataset used to train a YOLOv10-based vision model. The model is then deployed and tested on an edge platform in underwater robotics, achieving a state-of-the-art 0.657 mAP@50 for oyster detection on the Aqua2 platform.

Robots are increasingly envisioned to interact in real-world scenarios, where they must continuously adapt to new situations. To detect and grasp novel objects, zero-shot pose estimators determine poses without prior knowledge. Recently, vision language models (VLMs) have shown considerable advances in robotics applications by establishing an understanding between language input and image input. In our work, we take advantage of VLMs zero-shot capabilities and translate this ability to 6D object pose estimation. We propose a novel framework for promptable zero-shot 6D object pose estimation using language embeddings. The idea is to derive a coarse location of an object based on the relevancy map of a language-embedded NeRF reconstruction and to compute the pose estimate with a point cloud registration method. Additionally, we provide an analysis of LERF's suitability for open-set object pose estimation. We examine hyperparameters, such as activation thresholds for relevancy maps and investigate the zero-shot capabilities on an instance- and category-level. Furthermore, we plan to conduct robotic grasping experiments in a real-world setting.

The L\'evi-Civita connection of a Riemannian manifold is a metric (compatible) linear connection, uniquely determined by its vanishing torsion. It is extremal in the sense that it has minimal torsion at each point. We can extend this idea to more general spaces with more general (not necessarily quadratic) indicatrix hypersurfaces in the tangent spaces. Here, the existence of compatible linear connections on the base manifold is not guaranteed anymore, which needs to be addressed along with the intrinsic characterization of the extremal one. The first step is to provide the Riemann metrizability of the compatible linear connections. This Riemannian environment establishes a one-to-one correspondence between linear connections and their torsion tensors, also giving a way of measuring the length of the latter. The second step is to solve a hybrid conditional extremum problem at each point of the base manifold, all of whose constraint equations (compatibility equations) involve functions defined on the indicatrix hypersurface. The objective function to be minimized is a quadratic squared norm function defined on the finite dimensional fiber (vector space) of the torsion tensor bundle. We express the solution by using the method of Lagrange multipliers on function spaces point by point, we present a necessary and sufficient condition of the solvability and the solution is also given in terms of intrinsic quantities affecting the uniform size of the linear isometry groups of the indicatrices. This completes the description of differential geometric spaces admitting compatible linear connections on the base manifold, called generalized Berwald spaces, in Finsler geometry.

In the paper we investigate locally symmetric polynomial metrics in special cases of Riemannian and Finslerian surfaces. The Riemannian case will be presented by a collection of basic results (regularity of second root metrics) and formulas up to Gauss curvature. In case of Finslerian surfaces we formulate necessary and sufficient conditions for a locally symmetric fourth root metric in 2D to be positive definite. They are given in terms of the coefficients of the polynomial metric to make checking the positive definiteness as simple and direct as possible. Explicit examples are also presented. The situation is more complicated in case of spaces of dimension more than two. Some necessary conditions and an explicit example are given for a positive definite locally symmetric polynomial metric in 3D. Computations are supported by the MAPLE mathematics software (LinearAlgebra).

In a simple, undirected graph G, an edge 2-coloring is a coloring of the edges such that no vertex is incident to edges with more than 2 distinct colors. The problem maximum edge 2-coloring (ME2C) is to find an edge 2-coloring in a graph G with the goal to maximize the number of colors. For a relevant graph class, ME2C models anti-Ramsey numbers and it was considered in network applications. For the problem a 2-approximation algorithm is known, and if the input graph has a perfect matching, the same algorithm has been shown to have a performance guarantee of 5/3. It is known that ME2C is APX-hard and that it is UG-hard to obtain an approximation ratio better than 1.5. We show that if the input graph has a perfect matching, there is a polynomial time 1.625-approximation and if the graph is claw-free or if the maximum degree of the input graph is at most three (i.e., the graph is subcubic), there is a polynomial time 1.5-approximation algorithm for ME2C, Comment: 20 pages, 6 figures, preliminary version published at WAOA 2023

Estimating 2D-3D correspondences between RGB images and 3D space is a fundamental problem in 6D object pose estimation. Recent pose estimators use dense correspondence maps and Point-to-Point algorithms to estimate object poses. The accuracy of pose estimation depends heavily on the quality of the dense correspondence maps and their ability to withstand occlusion, clutter, and challenging material properties. Currently, dense correspondence maps are estimated using image-to-image translation models based on GANs, Autoencoders, or direct regression models. However, recent advancements in image-to-image translation have led to diffusion models being the superior choice when evaluated on benchmarking datasets. In this study, we compare image-to-image translation networks based on GANs and diffusion models for the downstream task of 6D object pose estimation. Our results demonstrate that the diffusion-based image-to-image translation model outperforms the GAN, revealing potential for further improvements in 6D object pose estimation models., Comment: Submitted to the First Austrian Symposium on AI, Robotics, and Vision 2024

Recent advances in machine learning have greatly benefited object detection and 6D pose estimation. However, textureless and metallic objects still pose a significant challenge due to few visual cues and the texture bias of CNNs. To address his issue, we propose a strategy for inducing a shape bias to CNN training. In particular, by randomizing textures applied to object surfaces during data rendering, we create training data without consistent textural cues. This methodology allows for seamless integration into existing data rendering engines, and results in negligible computational overhead for data rendering and network training. Our findings demonstrate that the shape bias we induce via randomized texturing, improves over existing approaches using style transfer. We evaluate with three detectors and two pose estimators. For the most recent object detector and for pose estimation in general, estimation accuracy improves for textureless and metallic objects. Additionally we show that our approach increases the pose estimation accuracy in the presence of image noise and strong illumination changes. Code and datasets are publicly available at github.com/hoenigpeter/randomized_texturing.

Real-time applications for autonomous operations depend largely on fast and robust vision-based localization systems. Since image processing tasks require processing large amounts of data, the computational resources often limit the performance of other processes. To overcome this limitation, traditional marker-based localization systems are widely used since they are easy to integrate and achieve reliable accuracy. However, classical marker-based localization systems significantly depend on standard cameras with low frame rates, which often lack accuracy due to motion blur. In contrast, event-based cameras provide high temporal resolution and a high dynamic range, which can be utilized for fast localization tasks, even under challenging visual conditions. This paper proposes a simple but effective event-based pose estimation system using active LED markers (ALM) for fast and accurate pose estimation. The proposed algorithm is able to operate in real time with a latency below \SI{0.5}{\milli\second} while maintaining output rates of \SI{3}{\kilo \hertz}. Experimental results in static and dynamic scenarios are presented to demonstrate the performance of the proposed approach in terms of computational speed and absolute accuracy, using the OptiTrack system as the basis for measurement., Comment: 14 pages, 12 figures, this paper has been accepted to WACV 2024

As robotic systems increasingly encounter complex and unconstrained real-world scenarios, there is a demand to recognize diverse objects. The state-of-the-art 6D object pose estimation methods rely on object-specific training and therefore do not generalize to unseen objects. Recent novel object pose estimation methods are solving this issue using task-specific fine-tuned CNNs for deep template matching. This adaptation for pose estimation still requires expensive data rendering and training procedures. MegaPose for example is trained on a dataset consisting of two million images showing 20,000 different objects to reach such generalization capabilities. To overcome this shortcoming we introduce ZS6D, for zero-shot novel object 6D pose estimation. Visual descriptors, extracted using pre-trained Vision Transformers (ViT), are used for matching rendered templates against query images of objects and for establishing local correspondences. These local correspondences enable deriving geometric correspondences and are used for estimating the object's 6D pose with RANSAC-based PnP. This approach showcases that the image descriptors extracted by pre-trained ViTs are well-suited to achieve a notable improvement over two state-of-the-art novel object 6D pose estimation methods, without the need for task-specific fine-tuning. Experiments are performed on LMO, YCBV, and TLESS. In comparison to one of the two methods we improve the Average Recall on all three datasets and compared to the second method we improve on two datasets.

Tracking an object's 6D pose, while either the object itself or the observing camera is moving, is important for many robotics and augmented reality applications. While exploiting temporal priors eases this problem, object-specific knowledge is required to recover when tracking is lost. Under the tight time constraints of the tracking task, RGB(D)-based methods are often conceptionally complex or rely on heuristic motion models. In comparison, we propose to simplify object tracking to a reinforced point cloud (depth only) alignment task. This allows us to train a streamlined approach from scratch with limited amounts of sparse 3D point clouds, compared to the large datasets of diverse RGBD sequences required in previous works. We incorporate temporal frame-to-frame registration with object-based recovery by frame-to-model refinement using a reinforcement learning (RL) agent that jointly solves for both objectives. We also show that the RL agent's uncertainty and a rendering-based mask propagation are effective reinitialization triggers., Comment: International Conference on Computer Vision Systems (ICVS) 2023

Object pose estimation is a core perception task that enables, for example, object grasping and scene understanding. The widely available, inexpensive and high-resolution RGB sensors and CNNs that allow for fast inference based on this modality make monocular approaches especially well suited for robotics applications. We observe that previous surveys on object pose estimation establish the state of the art for varying modalities, single- and multi-view settings, and datasets and metrics that consider a multitude of applications. We argue, however, that those works' broad scope hinders the identification of open challenges that are specific to monocular approaches and the derivation of promising future challenges for their application in robotics. By providing a unified view on recent publications from both robotics and computer vision, we find that occlusion handling, novel pose representations, and formalizing and improving category-level pose estimation are still fundamental challenges that are highly relevant for robotics. Moreover, to further improve robotic performance, large object sets, novel objects, refractive materials, and uncertainty estimates are central, largely unsolved open challenges. In order to address them, ontological reasoning, deformability handling, scene-level reasoning, realistic datasets, and the ecological footprint of algorithms need to be improved., Comment: Accepted for publication in Transactions on Robotics: https://ieeexplore.ieee.org/document/10609560 arXiv admin note: substantial text overlap with arXiv:2302.11827

Object pose estimation is important for object manipulation and scene understanding. In order to improve the general applicability of pose estimators, recent research focuses on providing estimates for novel objects, that is objects unseen during training. Such works use deep template matching strategies to retrieve the closest template connected to a query image. This template retrieval implicitly provides object class and pose. Despite the recent success and improvements of Vision Transformers over CNNs for many vision tasks, the state of the art uses CNN-based approaches for novel object pose estimation. This work evaluates and demonstrates the differences between self-supervised CNNs and Vision Transformers for deep template matching. In detail, both types of approaches are trained using contrastive learning to match training images against rendered templates of isolated objects. At test time, such templates are matched against query images of known and novel objects under challenging settings, such as clutter, occlusion and object symmetries, using masked cosine similarity. The presented results not only demonstrate that Vision Transformers improve in matching accuracy over CNNs, but also that for some cases pre-trained Vision Transformers do not need fine-tuning to do so. Furthermore, we highlight the differences in optimization and network architecture when comparing these two types of network for deep template matching.

A distance mean function measures the average distance of points from the elements of a given set of points (focal set) in the space. The level sets of a distance mean function are called generalized conics. In case of infinite focal points the average distance is typically given by integration over the focal set. The paper contains a survey on the applications of taxicab distance mean functions and generalized conics' theory in geometric tomography: bisection of the focal set and reconstruction problems by coordinate X-rays. The theoretical results are illustrated by implementations in Maple, methods and examples as well., Comment: 26 pages, 6 figures, the paper is based on the plenary lecture presented at Meeting on Tomography and Applications (Discrete Tomography, Neuroscience and Image Reconstruction) 16th Edition, IN MEMORIAM OF CARLA PERI, 2 - 4 May 2022, Mathematics Department, Politecnico di Milano, Milano, Italy

Object pose estimation is a non-trivial task that enables robotic manipulation, bin picking, augmented reality, and scene understanding, to name a few use cases. Monocular object pose estimation gained considerable momentum with the rise of high-performing deep learning-based solutions and is particularly interesting for the community since sensors are inexpensive and inference is fast. Prior works establish the comprehensive state of the art for diverse pose estimation problems. Their broad scopes make it difficult to identify promising future directions. We narrow down the scope to the problem of single-shot monocular 6D object pose estimation, which is commonly used in robotics, and thus are able to identify such trends. By reviewing recent publications in robotics and computer vision, the state of the art is established at the union of both fields. Following that, we identify promising research directions in order to help researchers to formulate relevant research ideas and effectively advance the state of the art. Findings include that methods are sophisticated enough to overcome the domain shift and that occlusion handling is a fundamental challenge. We also highlight problems such as novel object pose estimation and challenging materials handling as central challenges to advance robotics., Comment: Revised version in the making

Transparent objects are common in day-to-day life and hence find many applications that require robot grasping. Many solutions toward object grasping exist for non-transparent objects. However, due to the unique visual properties of transparent objects, standard 3D sensors produce noisy or distorted measurements. Modern approaches tackle this problem by either refining the noisy depth measurements or using some intermediate representation of the depth. Towards this, we study deep learning 6D pose estimation from RGB images only for transparent object grasping. To train and test the suitability of RGB-based object pose estimation, we construct a dataset of RGB-only images with 6D pose annotations. The experiments demonstrate the effectiveness of RGB image space for grasping transparent objects.

State-of-the-art object pose estimation handles multiple instances in a test image by using multi-model formulations: detection as a first stage and then separately trained networks per object for 2D-3D geometric correspondence prediction as a second stage. Poses are subsequently estimated using the Perspective-n-Points algorithm at runtime. Unfortunately, multi-model formulations are slow and do not scale well with the number of object instances involved. Recent approaches show that direct 6D object pose estimation is feasible when derived from the aforementioned geometric correspondences. We present an approach that learns an intermediate geometric representation of multiple objects to directly regress 6D poses of all instances in a test image. The inherent end-to-end trainability overcomes the requirement of separately processing individual object instances. By calculating the mutual Intersection-over-Unions, pose hypotheses are clustered into distinct instances, which achieves negligible runtime overhead with respect to the number of object instances. Results on multiple challenging standard datasets show that the pose estimation performance is superior to single-model state-of-the-art approaches despite being more than ~35 times faster. We additionally provide an analysis showing real-time applicability (>24 fps) for images where more than 90 object instances are present. Further results show the advantage of supervising geometric-correspondence-based object pose estimation with the 6D pose.

This paper presents BURG-Toolkit, a set of open-source tools for Benchmarking and Understanding Robotic Grasping. Our tools allow researchers to: (1) create virtual scenes for generating training data and performing grasping in simulation; (2) recreate the scene by arranging the corresponding objects accurately in the physical world for real robot experiments, supporting an analysis of the sim-to-real gap; and (3) share the scenes with other researchers to foster comparability and reproducibility of experimental results. We explain how to use our tools by describing some potential use cases. We further provide proof-of-concept experimental results quantifying the sim-to-real gap for robot grasping in some example scenes. The tools are available at: https://mrudorfer.github.io/burg-toolkit/, Comment: presented at ICRA 2022 Workshop on Releasing Robots into the Wild: Simulations, Benchmarks, and Deployment. Project page: https://mrudorfer.github.io/burg-toolkit/

Finsler metrics are direct generalizations of Riemannian metrics such that the quadratic Riemannian indicatrices in the tangent spaces of a manifold are replaced by more general convex bodies as unit spheres. A linear connection on the base manifold is called compatible with the Finsler metric if the induced parallel transports preserve the Finslerian length of tangent vectors. Finsler manifolds admitting compatible linear connections are called generalized Berwald manifolds. Compatible linear connections are the solutions of the so-called compatibility equations containing the torsion components as unknowns. Although there are some theoretical results for the solvability of the compatibility equations (monochromatic Finsler metrics \cite{BM}, extremal compatible linear connections, algorithmic solutions \cite{V14}), it is very hard to solve in general because compatible linear connections may or may not exist on a Finsler manifold and may or may not be unique. Therefore special cases are of special interest. One of them is the case of the so-called semi-symmetric compatible linear connection with decomposable torsion tensor. It is proved \cite{V10} (see also \cite{V11}) that such a compatible linear connection must be uniquely determined. The original proof is based on averaging in the sense that the 1-form in the decomposition of the torsion tensor can be expressed by integrating differential forms on the tangent manifold over the Finslerian indicatrices. The integral formulas are very difficult to compute in practice. We present a new proof for the unicity result by using linear algebra and some basic facts about convex bodies and an explicit formula for the solution without integration. Necessary conditions of the solvability are also formulated in terms of intrinsic equations without unknown quantities. They are sufficient if and only if the solution depends only on the position.

A fundamental variant of the classical traveling salesman problem (TSP) is the so-called multiple TSP (mTSP), where a set of $m$ salesmen jointly visit all cities from a set of $n$ cities. The mTSP models many important real-life applications, in particular for vehicle routing problems. An extensive survey by Bektas (Omega 34(3), 2006) lists a variety of heuristic and exact solution procedures for the mTSP, which quickly solve particular problem instances. In this work we consider a further generalization of mTSP, the many-visits mTSP, where each city $v$ has a request $r(v)$ of how many times it should be visited by the salesmen. This problem opens up new real-life applications such as aircraft sequencing, while at the same time it poses several computational challenges. We provide multiple efficient approximation algorithms for important variants of the many-visits mTSP, which are guaranteed to quickly compute high-quality solutions for all problem instances., Comment: 25 pages, 3 figures

Observational noise, inaccurate segmentation and ambiguity due to symmetry and occlusion lead to inaccurate object pose estimates. While depth- and RGB-based pose refinement approaches increase the accuracy of the resulting pose estimates, they are susceptible to ambiguity in the observation as they consider visual alignment. We propose to leverage the fact that we often observe static, rigid scenes. Thus, the objects therein need to be under physically plausible poses. We show that considering plausibility reduces ambiguity and, in consequence, allows poses to be more accurately predicted in cluttered environments. To this end, we extend a recent RL-based registration approach towards iterative refinement of object poses. Experiments on the LINEMOD and YCB-VIDEO datasets demonstrate the state-of-the-art performance of our depth-based refinement approach., Comment: IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2022

An equidistant polytope is a special equidistant set in the space $\mathbb{R}^n$ all of whose boundary points have equal distances from two finite systems of points. Since one of the finite systems of the given points is required to be in the interior of the convex hull of the other one we can speak about inner and outer focal points of the equidistant polytope. It is of type $(q, p)$, where $q$ is the number of the outer focal points and $p$ is the number of the inner focal points. The equidistancy is the generalization of convexity because a convex polytope can be given as an equidistant polytope of type $(q, 1)$, where $q\geq n+1$. In the paper we present some general results about the basic properties of the equidistant polytopes: convex components, graph representations, connectedness, correspondence to the Voronoi decomposition of the space etc. Especially, we are interested in equidistant polytopes of dimension $2$ (equidistant polygons). Equidistant polygons of type $(3,2)$ will be characterized in terms of a constructive (ruler-and-compass) process to recognize them. In general they are pentagons with exactly two concave angles such that the vertices, where the concave angles appear at, are joined by an inner diagonal related to the adjacent sides of the polygon in a special way via the three reflection theorem for concurrent lines. The last section is devoted to some special arrangements of the focal points to get the concave quadrangles as equidistant polygons of type $(3,2)$.

Motion and dynamic environments, especially under challenging lighting conditions, are still an open issue for robust robotic applications. In this paper, we propose an end-to-end pipeline for real-time, low latency, 6 degrees-of-freedom pose estimation of fiducial markers. Instead of achieving a pose estimation through a conventional frame-based approach, we employ the high-speed abilities of event-based sensors to directly refine the spatial transformation, using consecutive events. Furthermore, we introduce a novel two-way verification process for detecting tracking errors by backtracking the estimated pose, allowing us to evaluate the quality of our tracking. This approach allows us to achieve pose estimation at a rate up to 156~kHz, while only relying on CPU resources. The average end-to-end latency of our method is 3~ms. Experimental results demonstrate outstanding potential for robotic tasks, such as visual servoing in fast action-perception loops.

We report on a set of laboratory experiments to investigate the effect of Arctic warming on the amplitude and drift speed of the mid-latitude jet stream. Our results show that a progressive decrease of the meridional temperature difference 1) slows down the eastward propagation of the jet stream, 2) complexifies its structure, and 3) increases the frequency of extreme events. Extreme events and temperature variability show a clear trend in relation to the Arctic warming only at latitudes influenced by the jet stream, whilst such trend reverses in the equatorial region south of the subtropical jet. Despite missing land-sea contrast in the laboratory model, we find similar trends of temperature variability and extreme events in the experimental data and the National Centers for Environmental Prediction (NCEP) reanalysis data. Moreover, our data qualitatively confirm the decrease in temperature variability due to the meridional temperature gradient weakening (which has been proposed recently based on proxy data). Probability distributions are weakly sensitive to changes in the temperature gradient, which can be explained by recent findings using quasigeostrophic models., Comment: 14 pages, 13 figures

A linear connection on a Finsler manifold is called compatible to the Finsler function if its parallel transports preserve the Finslerian length of tangent vectors. Generalized Berwald manifolds are Finsler manifolds equipped with a compatible linear connection. In the paper we present a general and intrinsic method to characterize the compatible linear connections on a Finsler manifold of dimension three. We prove that if a compatible linear connection is not unique then the indicatrices must be Euclidean surfaces of revolution. The surplus freedom of choosing compatible linear connections is related to Euclidean symmetries. The unicity of the solution of the compatibility equations can be provided by some additional requirements. Following the idea in \cite{V14} we are also looking for the so-called extremal compatible linear connection minimizing the norm of its torsion at each point of the manifold.

Synthetic data generation has become essential in last years for feeding data-driven algorithms, which surpassed traditional techniques performance in almost every computer vision problem. Gathering and labelling the amount of data needed for these data-hungry models in the real world may become unfeasible and error-prone, while synthetic data give us the possibility of generating huge amounts of data with pixel-perfect annotations. However, most synthetic datasets lack from enough realism in their rendered images. In that context UnrealROX generation tool was presented in 2019, allowing to generate highly realistic data, at high resolutions and framerates, with an efficient pipeline based on Unreal Engine, a cutting-edge videogame engine. UnrealROX enabled robotic vision researchers to generate realistic and visually plausible data with full ground truth for a wide variety of problems such as class and instance semantic segmentation, object detection, depth estimation, visual grasping, and navigation. Nevertheless, its workflow was very tied to generate image sequences from a robotic on-board camera, making hard to generate data for other purposes. In this work, we present UnrealROX+, an improved version of UnrealROX where its decoupled and easy-to-use data acquisition system allows to quickly design and generate data in a much more flexible and customizable way. Moreover, it is packaged as an Unreal plug-in, which makes it more comfortable to use with already existing Unreal projects, and it also includes new features such as generating albedo or a Python API for interacting with the virtual environment from Deep Learning frameworks., Comment: Accepted at International Joint Conference on Neural Networks (IJCNN) 2021