GEPETTO Research Group

My topics of interest are currently collision-free motion planning accounting for system dynamics, the modeling of the natural human locomotion, digital mannequins and humanoid robots.

After ten years of work on nonholonomic system collision-free motion planning and control, I have decided to shift from wheels to legs. My current research interest is thus now vision-based motion planning and control for humanoid robots and by extension for other dynamic systems.

For the past ten years, I have been working on nonlinear control of wheeled robots and flying robots. My interest for sensor-based control, visual-servoing and sound perception, brought me to the study of neural processes related to multi-sensory and sensorimotor integration in primates, in collaboration with neuroscientists. At the same time I am studying today the problem humanoid control within the Gepetto Team.

I worked during several years in the field of motion planning for mobile robot (motion planning on rough terain, robust sensor based motion planning, coverage path planning). I am now involved in interactive motion planning via haptic feedback, motion planning for virtual mannequins collaboration and human motion.

My research focuses on modeling, numerical optimization and control algorithms for fast motions in robotics and biomechanics (such as walking, running, diving, juggling...). I am in particular interested in optimizing the stability and robustness of locomotion which leads to nondifferentiable optimization problems.

My research activities are concerned with visual servoing, and more specifically the integration of visual-servoing schemes into real robot applications. This research topic is at the intersection of the fields of robotics, automatic control, and computer vision. My main application field is currently the Humanoid robotics, as it causes serious challenges that can be found in many other robots domain.

My research interests are human-like walking pattern generation from motion capture data and real-time biped gait planning. The final goal is to achieve a humanoid robot that can go anywhere which human can go.

My main areas of interest are visual SLAM with detection and tracking of moving obstacles for Autonomous Navigation and non-linear, non-Gaussian filtering techniques.

I defended my thesis on rover localization in natural environments in July 2001. Since then, I've been wandering around here and here. Besides autonomous navigation, my main concern is software architecture for robots.

My PhD subject is multidisciplinary in which I study human motor control models (Neuroscience) for applications in controlling anthropomorphic systems (Robotics). Some methods relatives to this study are optimal control, visual servoing, feature extraction and machine learning.

I am working on motion planning for digital actors. I am particularly interested in planning algorithms for high dimensioned and strongly constrained systems.

The main subject of my thesis is about probabilistic algorithms for interactive motion planning in mechanical assembly tasks. My work is related to an industrial project called AMSI, dealing with "Product Lifecycle Management".

My thesis work is defined at the border of Neurosciences and Robotics. It concerns the modeling of the sensory transformations that allow the perception of "spatial constancy". Based on ectrophysiological studies in the visual cortex of monkey, my work aims at characterizing the invariant references that allow to construct a stable and unified representation of the scene and serve as a basis for the control of motion. In this way we are expecting to characterize basic principles that would allow to simplify the control o f humanoid robots.

My research work was focused mainly on the visual tracking using fuzzy logic and fuzzy sets. Within the Gepetto group, I have a completely new challenge and I will focus on Transfering Motion to a Humanoid Robot.

My current research tries to figure out how to produce a natural locomotor movement model in an environment including fixed and mobile obstacles by using an optimal control approach. After that, we will extend it to whole-body movements.

My thesis focuses on building vision guided models for locomotion for eventual application on humanoid robots. The idea is to study the link between vision and locomotion inferred from neuroscience models in humans. With a clear picture of this interaction we then evaluate its effectiveness in closing the loop between these two modalities in anthropomorphic robots.

My thesis concerns the generation of realistic human motion in order to help conceiving workplaces. This subject aims to make productivity more efficient, on a short term, and to reduce musculoskeletal disorders, on a long term. The main goal is to provide a realistic prediction of the motion, and eventually of certain efforts, required for a given manipulation task.

I am working on reactive imitation for humanoid robots. The idea is to explore possible classification of the trajectories generated by the control of the robot. And then, given a trajectory provided by a teacher and the classification, the robot choose reactively which controller to activate in order to perform an imitation.

My thesis work focuses on visual servoing and path planning. On one hand, visual servoing allows task definition which includes semantic information. On the other hand, path planning aims at generating collision free paths. The goal of my work would be to gather trajectory generation and semantic driven tasks in a global framework.

I worked previously on controlling robotic arms performing prioritized multi-tasks under constraints. My PhD thesis at LAAS focuses on motion planning for anthropomorphic systems, with application on humanoid robots and digital actors.

My research interests include motion and task planning including of many degree-of-freedom systems (humanoids and modular robots) including dynamics. I have been serving as one of co-directors of JRL (CNRS/AIST Joint French-Japanese Robotics Laboratory) located in LAAS-CNRS with Dr. Jean-Paul Laumond.

I have worked several years to develop a software environment for humanoid robots which is called OpenHRP (Open architecture Humanoid Robotics Platform). Current research interests include simulatenous whole body motion planning and 3D environment mapping.

I worked on motion planning for humanoid robots. Other topics of intrest included probabilistic robotics, machine learning and computer animation.

I started my work in the field of Robotics on Kinematic and Dynamic Modeling of Mobile Manipulators. As a PhD student, I worked on Humanoid Robot(HRP2) Motion Planning in 3D Environments.

My thesis was around two main subjects: motion planning for humanoid robot using probabilistic roadmap and pattern generator; Manipulation task planning for humanoid robots using pivoting technic.

My thesis focused on the optimisation of motion overlapping for task sequencing. Starting from a sequence of tasks given (e.g. the result of a planification), we wanted to optimize the movement of the robot by taking advantage of its redundancy in order to make it faster, smoother and more realistic.

The main subject of my thesis was modeling human movements using system identification theory. My interesting research areas are synthesizing human captured data and identifying nonlinear dynamic systems. I also was a member of the MRS group collaborating with Gepetto.

I worked on mechanical assembly problems in the context of industrial Product Lifecycle Management (PLM) as well as ergonomics. My work was supported by a collaborative research action with Kineo CAM.

I analyzed human locomotor trajectories from an optimal control perspective. Our approach emphasized the close relationship between the shape of locomotor paths in goal-directed movements and nonholonomic mobile robots.

My Ph.D. dealt with "Motion Planning applied to car axle design". The objective was to determine with precision the maximum clearing speed of classical automotive driving tests (VDA, curve braking, ...). The correct trajectory of the industrial car model was computed thanks to motion planning algorithms.

I have been working on motion planning algorithms for redundant mechanisms with a large number of degrees of freedom. These algorithms account for the geometry, kinematics and dynamics of the devices as well as the interaction with the environment. I defended my thesis in February, 2007 and left the team in May, 2007.

Engineer in Computer Science and Applied Mathematics from ENSEEIHT. My research deals with path planning and trajectory execution for non-holonomic systems.