Recent advances in AI have led to significant results in robotic learning, but skills like grasping remain partially solved. Many recent works exploit synthetic grasping datasets to learn to grasp unknown objects. However, those datasets were generated using simple grasp sampling methods using priors. Recently, Quality-Diversity (QD) algorithms have been proven to make grasp sampling significantly more efficient. In this work, we extend QDG-6DoF, a QD framework for generating object-centric grasps, to scale up the production of synthetic grasping datasets. We propose a data augmentation method that combines the transformation of object meshes with transfer learning from previous grasping repertoires. The conducted experiments show that this approach reduces the number of required evaluations per discovered robust grasp by up to 20 %. We used this approach to generate QDGset, a dataset of 6 DoF grasp poses that contains about 3.5 and 4.5 times more grasps and objects, respectively, than the previous state-of-the-art. Our method allows anyone to easily generate data, eventually contributing to a large-scale collaborative dataset of synthetic grasps.

Object handover is an important skill that we use daily when interacting with other humans. To deploy robots in collaborative setting, like houses, being able to receive and handing over objects safely and efficiently becomes a crucial skill. In this work, we demonstrate the use of Reinforcement Learning (RL) for dexterous object handover between two multi-finger hands. Key to this task is the use of a novel reward function based on dual quaternions to minimize the rotation distance, which outperforms other rotation representations such as Euler and rotation matrices. The robustness of the trained policy is experimentally evaluated by testing w.r.t. objects that are not included in the training distribution, and perturbations during the handover process. The results demonstrate that the trained policy successfully perform this task, achieving a total success rate of 94% in the best-case scenario after 100 experiments, thereby showing the robustness of our policy with novel objects. In addition, the best-case performance of the trained policy decreases by only 13.8% when the other robot moves during the handover, proving that our policy is also robust to this type of perturbation, which is common in real-world object handovers. Code and videos can be found here.

The task of grasping objects is a simple and routinely action for humans but it is complex for robots. To integrate robots into everyday tasks, they have to be equipped with capabilities human-like dexterity. In this line, we propose an analytic method, called GeoGraspEvo, to compute grasping points to be used by robotic hands with three, four or more fingers. Our proposal uses features computed from visible surface objects captured by a single RGBD image of a scene. Additionally, it uses as input some configurable kinematic parameters to be able to carry out the grasping depending on the hand morphology. The method compute grasping points with no training process.

LiDAR sensors are used for mapping tasks, LiDAR odometry or 3D environment reconstruction. Several of them count with a high number of vertical layers, which increase their price and prevents research groups from carrying out experiments and scientific advances. In this paper, we propose a method for augmenting point cloud data by bilinear interpolation in a Spherical Range Image. Our method improves others on the state-of-the-art by means of standard deviation filtering of the newly generated layers. The system operates at a frequency greater than 10 Hz for data interpolation up to 20 times. In addition, we present two applications for our approach such as LiDAR odometry and LiDAR-Camera fusion, obtaining better results than others that do not apply data augmentation. Finally we make available to the scientific community a package development on ROS (Robot Operating System). The code is available at https://github.com/EPVelasco/lidar-camera-fusion

Nowadays, robotics education programs are used to promote the skills of students in STEM content at different levels of education for a wide variety of studies, not just engineering. This paper describes an educational program of robotics, applied precisely to the laboratory sessions of a robotics initiation course taught in the first year of a degree in Robotic Engineering, as well as the results of the application of that program in different academic years. The educational program contemplates that the students, grouped in work teams, have to design, assemble, and program a small mobile robot, destined to a final competition in which all the teams participate.

Nowadays, Robotics is increasing its importance for the marine environment in our society. It allows to perform surveillance and data sampling tasks reducing the current cost of these tasks. This paper presents the design of a prototype of an Autonomous Surface Vessel (ASV) based on catamaran shape. It uses the screw theory for the propulsion system, which provides high manoeuvrability to the vessel. Furthermore, the parts of the ASV are designed to be printed by a RepRap 3D printer. This gives flexibility to check the performance of the vessel. Also the software control scheme of the vessel is presented.

This paper shows the state of art teleoperation and simulation systems and proposes some possible network architectures devoted to the development of such systems. A method for the optimization of robot tasks is also suggested. The application proposed can teleoperate a robot arm with five degrees of freedom through Internet by using a previous simulation system and visual feedback. This simulation can be accessed simultaneously by different workers. As a result, the number of robots required can be reduced on new applications or dangerous environments.