Projects.Toucan History

Hide minor edits - Show changes to output

Changed line 4 from:
(:Start:2014:)
to:
(:Start:20??:)
Changed line 8 from:
(:Website:phonestat2014.imag.fr/:)
to:
(:comment Website:???:)
Changed lines 11-12 from:
(:Coordinator:[[Profiles/SamsonA|+]]:)
(:Collaborators:[[Profiles/LetueF|+]], [[Profiles/BazzoliC|+]], [[Profiles/MartinezMJ
|+]]:)
to:
(:comment Coordinator:???:)
(:Collaborators:
[[Profiles/YcartB|+]]:)
Changed line 20 from:
Le Département Parole et Cognition du GIPSA-lab s'intéresse à mieux comprendre la production et la perception de la voix et de la parole : quels indices acoustiques, articulatoires, gestuels sont produits et perçus ? Comment la parole se développe-t-elle chez l'enfant ? A quel niveau se situe la dysfonction dans le cas de troubles de la voix ou de la parole, et comment prévenir ou rééduquer ces troubles ? Des données expérimentales sont recueillies chez des individus, avec des protocoles spécifiques (répétitions des expériences dans différentes conditions), qui impliquent une analyse statistique fine des données. Or actuellement, les données sont traitées par analyse de la variance, qui n'est pas toujours l'outil statistique adapté. L'idée phare de ce projet est donc de diffuser au sein de cette communauté des outils statistiques novateurs adaptés à leurs protocoles, en particulier des modèles à effets aléatoires, qui devraient permettre d'extraire de nouvelles informations des données.
to:
???
Changed line 1 from:
(:comment Acronym:Cancer et génomique:)
to:
(:comment Acronym:???:)
Changed lines 1-5 from:
(:Acronym:Toucan:)
(:comment FullName:???:)

(:Start:2008:)
(:End:2012:)
to:
(:comment Acronym:Cancer et génomique:)
(:FullName:Cancer et génomique:)

(:Start:2014:)
(:comment End:????:)
Changed lines 8-13 from:
(:comment Website:???:)
(:Funding:ANR:)
(:Budget:180000€:)
(:Collaborators:[[Profiles/AndreffN|+]], Adrien Bartoli, François Berry, Omar Ait Aider:)
(:Topics:[[!Vision]] based [[!Control]] of parallel robots:)
to:
(:Website:phonestat2014.imag.fr/:)
(:Funding:LabEx Toucan:)
(:comment Budget:???€:)
(:Coordinator:[[Profiles/SamsonA|+]]:)
(:Collaborators:[[Profiles/LetueF|+]], [[Profiles/BazzoliC|+]], [[Profiles/MartinezMJ|+]]:)
(:Topics:
:)
Added line 20:
Le Département Parole et Cognition du GIPSA-lab s'intéresse à mieux comprendre la production et la perception de la voix et de la parole : quels indices acoustiques, articulatoires, gestuels sont produits et perçus ? Comment la parole se développe-t-elle chez l'enfant ? A quel niveau se situe la dysfonction dans le cas de troubles de la voix ou de la parole, et comment prévenir ou rééduquer ces troubles ? Des données expérimentales sont recueillies chez des individus, avec des protocoles spécifiques (répétitions des expériences dans différentes conditions), qui impliquent une analyse statistique fine des données. Or actuellement, les données sont traitées par analyse de la variance, qui n'est pas toujours l'outil statistique adapté. L'idée phare de ce projet est donc de diffuser au sein de cette communauté des outils statistiques novateurs adaptés à leurs protocoles, en particulier des modèles à effets aléatoires, qui devraient permettre d'extraire de nouvelles informations des données.
Changed line 1 from:
(:Acronym:Virago:)
to:
(:Acronym:Toucan:)
Changed line 7 from:
(:comment Logo:Virago.jpg:)
to:
(:comment Logo:logo.jpg:)
Deleted lines 18-20:
This project is part of a larger move to form a young and efficient team dedicated to the modeling, identification and control of parallel robots by vision. Modeling, identification and control of serial robots is decades old, as well as the use of vision in this context. When it comes to parallel robots, the litterature becomes much more dedicated to the design and modeling of such robots, with some scattered works on control and identification using conventional techniques. Furthermore, vision was seldom used for parallel robots. This coupling between vision and parallel robots started at LASMEA during fall 2000 as a « joint venture » with the mechanical engineering group (LaMI) in ClermontFerrand.

Since that time, it has produced several original results initially dedicated to vision-based kinematic calibration of parallel robots, published in international conferences and journals, both in the robotic and in the mechanical engineering communities. Since two years now, our research is being widened to address dynamic identification and control problems for parallel robots. As far as kinematic control is concerned, we noticed that much of the control solutions are outdated since they try to apply « as is » the methods coming from serial robotics. Indeed, either control is made in the joint space, which is an error since it does not take into account the kinematic chain closure constraint (or at least not properly), or control is planned to be made in the Cartesian space using joint sensing, which can not be effectively applied without solving for the theoretically hard Forward Kinematic Problem. We discovered that using vision, the control could be simplified since it relieves Cartesian control from the Forward Kinematic Problem. The reason for that is that the state of a parallel robot is its end-effector pose and not its joint configuration, as opposed to the serial case. As far as dynamic control is concerned now, we feel that the same errors are being made. Indeed, dynamic control differs from kinematic control since it does not neglect phenomena such as inertia or Coriolis forces (among others). It thus allows for higher robot speed, which is what parallel robots aim at. However, as far as we know, dynamic control of parallel robots is essentially made using joint sensors. Since the state of a parallel robot is the end effector pose, it should be wiser to perform control based on the sensing of the latter, which can be made by vision, but this has never been addressed before and raises the problem of the speed of the vision sensor (around 1kHz). Notice that this frequency is not reachable with uptodate commercial laser tracker systems or GPSlike systems.
Changed line 11 from:
(:Collaborators:[[Profiles/AndreffN|+]],Adrien Bartoli, François Berry, Omar Ait Aider:)
to:
(:Collaborators:[[Profiles/AndreffN|+]], Adrien Bartoli, François Berry, Omar Ait Aider:)
Changed lines 11-12 from:
(:Leaders:[[Profiles/AndreffN|+]]:)
(:Collaborators:
Adrien Bartoli, François Berry, Omar Ait Aider:)
to:
(:Collaborators:[[Profiles/AndreffN|+]],Adrien Bartoli, François Berry, Omar Ait Aider:)
Changed line 22 from:
Since that time, it has produced several original results initially dedicated to vision-based kinematic calibration of parallel robots, published in international conferences and journals, both in the robotic and in the mechanical engineering communities. Since two years now, our research is being widened to address dynamic identification and control problems for parallel robots. As far as kinematic control is concerned, we noticed that much of the control solutions are outdated since they try to apply « as is » the methods coming from serial robotics. Indeed, either control is made in the joint space, which is an error since it does not take into account the kinematic chain closure constraint (or at least not properly), or control is planned to be made in the Cartesian space using joint sensing, which can not be effectively applied without solving for the theoretically hard Forward Kinematic Problem. We discovered that using vision, the control could be simplified since it relieves Cartesian control from the Forward Kinematic Problem. The reason for that is that the state of a parallel robot is its end-effector pose and not its joint configuration, as opposed to the serial case. As far as dynamic control is concerned now, we feel that the same errors are being made. Indeed, dynamic control differs from kinematic control since it does not neglect phenomena such as inertia or Coriolis forces (among others). It thus allows for higher robot speed, which is what parallel robots aim at. However, as far as we know, dynamic control of parallel robots is essentially made using joint sensors. Since the state of a parallel robot is the endeffector pose, it should be wiser to perform control based on the sensing of the latter, which can be made by vision, but this has never been addressed before and raises the problem of the speed of the vision sensor (around 1kHz). Notice that this frequency is not reachable with uptodate commercial laser tracker systems or GPSlike systems.
to:
Since that time, it has produced several original results initially dedicated to vision-based kinematic calibration of parallel robots, published in international conferences and journals, both in the robotic and in the mechanical engineering communities. Since two years now, our research is being widened to address dynamic identification and control problems for parallel robots. As far as kinematic control is concerned, we noticed that much of the control solutions are outdated since they try to apply « as is » the methods coming from serial robotics. Indeed, either control is made in the joint space, which is an error since it does not take into account the kinematic chain closure constraint (or at least not properly), or control is planned to be made in the Cartesian space using joint sensing, which can not be effectively applied without solving for the theoretically hard Forward Kinematic Problem. We discovered that using vision, the control could be simplified since it relieves Cartesian control from the Forward Kinematic Problem. The reason for that is that the state of a parallel robot is its end-effector pose and not its joint configuration, as opposed to the serial case. As far as dynamic control is concerned now, we feel that the same errors are being made. Indeed, dynamic control differs from kinematic control since it does not neglect phenomena such as inertia or Coriolis forces (among others). It thus allows for higher robot speed, which is what parallel robots aim at. However, as far as we know, dynamic control of parallel robots is essentially made using joint sensors. Since the state of a parallel robot is the end effector pose, it should be wiser to perform control based on the sensing of the latter, which can be made by vision, but this has never been addressed before and raises the problem of the speed of the vision sensor (around 1kHz). Notice that this frequency is not reachable with uptodate commercial laser tracker systems or GPSlike systems.
Deleted lines 8-9:
(:comment Logo:Logo:)
(:comment Website:Website:)
Changed line 2 from:
(:FullName:???:)
to:
(:comment FullName:???:)
Deleted line 15:
[[<<]]
Added line 16:
[[<<]]
Added line 19:
----
Changed line 19 from:
!!Description
to:
!!!Description
Added lines 18-19:

!!Description
Changed lines 17-21 from:
(:title {$:Acronym}:)
to:
(:title {$:Acronym}:)

This project is part of a larger move to form a young and efficient team dedicated to the modeling, identification and control of parallel robots by vision. Modeling, identification and control of serial robots is decades old, as well as the use of vision in this context. When it comes to parallel robots, the litterature becomes much more dedicated to the design and modeling of such robots, with some scattered works on control and identification using conventional techniques. Furthermore, vision was seldom used for parallel robots. This coupling between vision and parallel robots started at LASMEA during fall 2000 as a « joint venture » with the mechanical engineering group (LaMI) in ClermontFerrand.

Since that time, it has produced several original results initially dedicated to vision-based kinematic calibration of parallel robots, published in international conferences and journals, both in the robotic and in the mechanical engineering communities. Since two years now, our research is being widened to address dynamic identification and control problems for parallel robots. As far as kinematic control is concerned, we noticed that much of the control solutions are outdated since they try to apply « as is » the methods coming from serial robotics. Indeed, either control is made in the joint space, which is an error since it does not take into account the kinematic chain closure constraint (or at least not properly), or control is planned to be made in the Cartesian space using joint sensing, which can not be effectively applied without solving for the theoretically hard Forward Kinematic Problem. We discovered that using vision, the control could be simplified since it relieves Cartesian control from the Forward Kinematic Problem. The reason for that is that the state of a parallel robot is its end-effector pose and not its joint configuration, as opposed to the serial case. As far as dynamic control is concerned now, we feel that the same errors are being made. Indeed, dynamic control differs from kinematic control since it does not neglect phenomena such as inertia or Coriolis forces (among others). It thus allows for higher robot speed, which is what parallel robots aim at. However, as far as we know, dynamic control of parallel robots is essentially made using joint sensors. Since the state of a parallel robot is the endeffector pose, it should be wiser to perform control based on the sensing of the latter, which can be made by vision, but this has never been addressed before and raises the problem of the speed of the vision sensor (around 1kHz). Notice that this frequency is not reachable with uptodate commercial laser tracker systems or GPSlike systems.
Changed line 6 from:
(:Status:Finished:)
to:
(:Status:In progress:)
Changed lines 5-6 from:
(:comment End:2012:)
to:
(:End:2012:)
(:Status:Finished
:)
Changed line 12 from:
(:Leaders:[[Profiles/AndreffN]]:)
to:
(:Leaders:[[Profiles/AndreffN|+]]:)
Changed line 12 from:
(:Leaders:[[Profiles/AndreffN:)
to:
(:Leaders:[[Profiles/AndreffN]]:)
Added line 3:
Added lines 8-13:
(:comment Logo:Logo:)
(:comment Website:Website:)
(:Funding:ANR:)
(:Budget:180000€:)
(:Leaders:[[Profiles/AndreffN:)
(:Collaborators:Adrien Bartoli, François Berry, Omar Ait Aider:)
Added lines 1-9:
(:Acronym:Virago:)
(:FullName:???:)
(:Start:2008:)
(:comment End:2012:)
(:comment Logo:Virago.jpg:)
(:comment Website:???:)
(:Topics:[[!Vision]] based [[!Control]] of parallel robots:)

(:title {$:Acronym}:)