I am a researcher and project manager at Tecnalia, Donostia, in the Spanish basque country. I work in the Medical Robotics group, from the Health Division, as well as in the Advanced Manufacturing group of the Industry and Transport Division. I am involved in the development of technological solutions for physical Human Robot interaction, vision-based robotic manipulation, … I am also very interested in software architecture, within (or without) the ROS framework.
PhD in Computer Science, 2004
Université de Rennes I
Master of Research in Image and Artificial Intelligence, 2001
Université de Rennes I
Engineer Degree in Computer Science, 2001
INSA of Rennes
This article deals with the 2D image-based recognition of industrial parts. Methods based on histograms are well known and widely used, but it is hard to find the best combination of histograms, most distinctive for instance, for each situation and without a high user expertise. We proposed a descriptor subset selection technique that automatically selects the most appropriate descriptor combination, and that outperforms approach involving single descriptors. We have considered both backward and forward mechanisms. Furthermore, to recognize the industrial parts a supervised classification is used with the global descriptors as predictors. Several class approaches are compared. Given our application, the best results are obtained with the Support Vector Machine with a combination of descriptors increasing the F1 by 0.031 with respect to the best descriptor
This paper is related to the observation of human operator manipulating objects for teaching a robot to reproduce the action. Assuming the robotic system is equipped with basic manipulation skills, we focus here on the automatic segmentation of the observed manipulation, for extracting the relevant key frames in which the manipulation is best described. The segmentation method proposed is based on the instantaneous work, and presents the advantage of not depending on the force and pose sensing locations. The experimentations concern two different manipulation skills, sliding and folding. We demonstrate in different settings that such segmentation method is efficient.
Detection and description of keypoints from an image is a well-studied problem in Computer Vision. Some methods like SIFT, SURF or ORB are computationally really efficient. This paper proposes a solution for a particular case study on object recognition of industrial parts based on hierarchical classification. Reducing the number of instances leads to better performance, indeed, that is what the use of the hierarchical classification is looking for. We demonstrate that this method performs better than using just one method like ORB, SIFT or FREAK, despite being fairly slower.
This paper describes a tool for generating ROS packages and nodes. Compared to the relatively basic traditional package creation method, this tool can generate a whole node structure, including its life-cycle and the exposed interface to other ROS nodes. Following a separation of concerns, the developer only defines the interaction means in a XML file, and the tool provides the whole skeleton of the nodes, including the interface creation and management. This way, the developer can focus on his real added value, the implementation of the node logic. Compared to advanced node management frameworks proposed in literature, the tool proposed does not require the developer to understand and agree on complex high-level architecture models. The developer only has to select a template model, and to provide the desired interface to get the code generated. The package generation is made possible thanks to package templates, and we provide with the generator tool two templates for creating nodes either in C or Python. The user has also the possibility to design his own template, so that he can develop the one that best fits his needs and best practices. The package generator code is accessible on public repository hosting facilities.
Traditional minimally invasive robots provide to the surgeon an interface for controlling the tip of the endoscopic arm in Cartesian space. We proposed therefore a similar interface for the STIFF-FLOP robot. The direct control of the tip pose was provided by an inverse kinematics component, computing the appropriate STIFF-FLOP robot configuration. Due to the flexibility of the arm modules, we have organized the inverse kinematics into two layers. The first one handles the inverse kinematics in a generic way. It is based on a numerical estimation of the robot. This layer is generic in the sense that it can incorporate any module representation, as long as the module representation provides a forward kinematics mechanism. The second layer concerns the kinematic modeling of the flexible modules, and has to provide forward kinematics functionalities for the upper model. Instead of the standard constant curvature parameters, we are proposing two other representations, one using each module tip position, and the other one directly using the chamber lengths. The flexible modules are connected to a robotic arm through a rigid rod, to extend the operational space of the system. The robotic arm pose is encoded with an adaptation of the spherical coordinate system to ensure that the rod entering the human body respects the single insertion point constraint. By defining a forward kinematics for the rod pose, the external robot end effector is implicitly embedded into the general inverse kinematics scheme, so that the estimation of the flexible modules’ configurations and the pose of the robot end-effector are all computed together to follow the motion requests provided by the surgeon..