J. Irving Vasquez-Gomez

Gabriel O. Flores-Aquino, Jheison Duvier Díaz Ortega, Ricardo Yahir Almazan Arvizu et al.

In the last decade, autonomous navigation for roboticshas been leveraged by deep learning and other approachesbased on machine learning. These approaches have demon-strated significant advantages in robotics performance. Butthey have the disadvantage that they require a lot of data toinfer knowledge. In this paper, we present an algorithm forbuilding 2D maps with attributes that make them useful fortraining and testing machine-learning-based approaches.The maps are based on dungeons environments where sev-eral random rooms are built and then those rooms are con-nected. In addition, we provide a dataset with 10,000 mapsproduced by the proposed algorithm and a description withextensive information for algorithm evaluation. Such infor-mation includes validation of path existence, the best path,distances, among other attributes. We believe that thesemaps and their related information can be very useful forrobotics enthusiasts and researchers who want to test deeplearning approaches. The dataset is available athttps://github.com/gbriel21/map2D_dataSet.git

Gabriel O. Flores-Aquino, J. Irving Vasquez-Gomez, O. Octavio Gutierrez-Frias

Sampling-based motion planning algorithms are widely used in robotics because they are very effective in high-dimensional spaces. However, the success rate and quality of the solutions are determined by an adequate selection of their parameters such as the distance between states, the local planner, and the sampling distribution. For robots with large configuration spaces or dynamic restrictions, selecting these parameters is a challenging task. This paper proposes a method for improving the performance to a set of the most popular sampling-based algorithms, the Rapidly-exploring Random Trees (RRTs) by adjusting the sampling method. The idea is to replace the uniform probability density function (U-PDF) with a custom distribution (C-PDF) learned from previously successful queries in similar tasks. With a few samples, our method builds a custom distribution that allows the RRT to grow to promising states that will lead to a solution. We tested our method in several autonomous driving tasks such as parking maneuvers, obstacle clearance and under narrow passages scenarios. The results show that the proposed method outperforms the original RRT and several improved versions in terms of success rate, tree density and computation time. In addition, the proposed method requires a relatively small set of examples, unlike current deep learning techniques that require a vast amount of examples.

J. Irving Vasquez-Gomez, David Troncoso, Israel Becerra et al.

Automated three-dimensional (3D) object reconstruction is the task of building a geometric representation of a physical object by means of sensing its surface. Even though new single view reconstruction techniques can predict the surface, they lead to incomplete models, specially, for non commons objects such as antique objects or art sculptures. Therefore, to achieve the task's goals, it is essential to automatically determine the locations where the sensor will be placed so that the surface will be completely observed. This problem is known as the next-best-view problem. In this paper, we propose a data-driven approach to address the problem. The proposed approach trains a 3D convolutional neural network (3D CNN) with previous reconstructions in order to regress the \btxt{position of the} next-best-view. To the best of our knowledge, this is one of the first works that directly infers the next-best-view in a continuous space using a data-driven approach for the 3D object reconstruction task. We have validated the proposed approach making use of two groups of experiments. In the first group, several variants of the proposed architecture are analyzed. Predicted next-best-views were observed to be closely positioned to the ground truth. In the second group of experiments, the proposed approach is requested to reconstruct several unseen objects, namely, objects not considered by the 3D CNN during training nor validation. Coverage percentages of up to 90 \% were observed. With respect to current state-of-the-art methods, the proposed approach improves the performance of previous next-best-view classification approaches and it is quite fast in running time (3 frames per second), given that it does not compute the expensive ray tracing required by previous information metrics.

Miguel Mendoza, J. Irving Vasquez-Gomez, Hind Taud et al.

Motivated by the advances in 3D sensing technology and the spreading of low-cost robotic platforms, 3D object reconstruction has become a common task in many areas. Nevertheless, the selection of the optimal sensor pose that maximizes the reconstructed surface is a problem that remains open. It is known in the literature as the next-best-view planning problem. In this paper, we propose a novel next-best-view planning scheme based on supervised deep learning. The scheme contains an algorithm for automatic generation of datasets and an original three-dimensional convolutional neural network (3D-CNN) used to learn the next-best-view. Unlike previous work where the problem is addressed as a search, the trained 3D-CNN directly predicts the sensor pose. We present a comparison of the proposed network against a similar net, and we present several experiments of the reconstruction of unknown objects validating the effectiveness of the proposed scheme.