with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of th... more with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of the resulting identification approach is tested and evaluated on a simulated benchmark hybrid system.
Atlas home || Conferences | Abstracts | about Atlas Mathematical Problems in Engineering, Aerospa... more Atlas home || Conferences | Abstracts | about Atlas Mathematical Problems in Engineering, Aerospace and Sciences June 25-27, 2008 University of Genoa, Italy Genoa, Italy. Organizers General Organizer and Chair: Seenith Sivasundaram, USA; Local organizer and Chair: Marcello Sanguineti, Italy. Conference Homepage. Abstracts. Rakhimov Abdumalik Summability of spectral expansions of distributions Lidia Aceto On the relations between block-BVMs and Runge-Kutta collocation methods. ...
This paper suggests an improved method for predictive control of hybrid systems with mixed inputs... more This paper suggests an improved method for predictive control of hybrid systems with mixed inputs. The algorithm takes into account the real nonlinear system as a model of a hybrid system, which is based on building a tree of evolution. Where the branch & bound (B&B) technique is applied for discrete controls in which an embedded nonlinear programming approach (Pattern search) is associated with each node of the tree in order to provide the continuous controls and explore the tree. Once the whole nodes of the tree are explored, the corresponding input is exploited to the system and the procedure is repeated. The performance of the resulting predictive control system is demonstrated on a motorboat simulation case study.
The Matlab/Octave programs on this page are free for academic use. Most of the source codes are i... more The Matlab/Octave programs on this page are free for academic use. Most of the source codes are implementations of the algorithms described in the book "Robot Programming by Demonstration: A Probabilistic Approach". ... Please provide references to the authors' publications in works using parts of these codes. ... The GMM-GMR sourcecode is also available in C/C++ (platform independent) at Sourceforge.net. ... Unzip the file and run 'demo1' in Matlab. ... Calinon, S., Pistillo, A. and Caldwell, DG (2011) Encoding the time and space constraints ...
In this paper we compare two recently proposed algorithms for online identification of hybrid sys... more In this paper we compare two recently proposed algorithms for online identification of hybrid systems. We consider the adaptive growing and pruning Radial Basis Function (RBF) neural network based and the potential fuzzy clustering based procedures. Specific behaviors of the procedures are pointed out, using a well known two dimensional example.
In this paper an innovative method for 1 verification of signature using parametric features base... more In this paper an innovative method for 1 verification of signature using parametric features based on optimal threshold selection is proposed. For each signature, 62 parametric feature are derived from horizontal place, x(t), vertical place, y(t) and pen down and up signals which are obtained from a digitizer plane. The weighted distance between each feature of a signatories and the related reference features is compared to a suitable threshold value and then the feature is accepted or not. The number of the accepted features for a person is then compared to another threshold, which has a suitable value for each signature, and then the signature will be verified or rejected. In this research, 1500 original signatures from 30 person and 600 forgery signatures are used. For each person, 30 genuine and 10 forgery signatures are considered for training of the algorithm and the rest are used in testing and validation. It is shown in the results that there is 0.67% false rejection ratio and 0.67% false acceptation ratio for the training set and a 2.68% and 1.99% for the testing set, respectively.
"Robot learning from demonstrations requires the
robot to learn and adapt movements to new situa... more "Robot learning from demonstrations requires the
robot to learn and adapt movements to new situations, often
characterized by position and orientation of objects or landmarks in the robot’s environment. In the task-parameterized
Gaussian mixture model framework, the movements are considered to be modulated with respect to a set of candidate
frames of reference (coordinate systems) attached to a set of
objects in the robot workspace. Following a similar approach,
this paper addresses the problem of having missing candidate
frames during the demonstrations and reproductions, which
can happen in various situations such as visual occlusion, sensor
unavailability, or tasks with a variable number of descriptive
features. We study this problem with a dust sweeping task in
which the robot requires to consider a variable amount of dust
areas to clean for each reproduction trial."
In contrast to standard industrial robots designed to work in predefined factory settings, humano... more In contrast to standard industrial robots designed to work in predefined factory settings, humanoids are designed to share our space without modifying our environment. These robots unveil new collaboration and social interaction perspectives, but these new environments also expose the robots to various sources of perturbation and unpredictable situations. Since the range of tasks that the humanoids can carry is infinite, it is not possible to provide a predefined database of tasks. For a widespread use in homes and offices, a key requirement will be to provide user-friendly ways of programming the robot to let it acquire new skills and adapt existing skills to new situations. Robot learning by imitation offers a promising route to transfer and refine skills from demonstration by non-expert users.
Gestures are characterized by intermediary or
final landmarks (real or virtual) in task space or... more Gestures are characterized by intermediary or
final landmarks (real or virtual) in task space or joint space
that can change during the course of the motion, and that
are described by varying accuracy and correlation constraints.
Generalizing these trajectories in robot learning by imitation
is challenging, because of the small number of demonstrations
provided by the user. We present an approach to statistically
encode movements in a task-parameterized mixture model,
and derive an expectation-maximization (EM) algorithm to
train it. The model automatically extracts the relevance of
candidate coordinate systems during the task, and exploits this
information during reproduction to adapt the movement in
real-time to changing position and orientation of landmarks or
objects. The approach is tested with a robotic arm learning to
roll out a pizza dough. It is compared to three categories of taskparameterized
models: 1) Gaussian process regression (GPR)
with a trajectory models database; 2) Multi-streams approach
with models trained in several frames of reference; and 3)
Parametric Gaussian mixture model (PGMM) modulating the
Gaussian centers with the task parameters. We show that the
extrapolation capability of the proposed approach outperforms
existing methods, by extracting the local structures of the task
instead of relying on interpolation principles.
with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of th... more with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of the resulting identification approach is tested and evaluated on a simulated benchmark hybrid system.
This paper suggests an improved method for
predictive control of hybrid systems with mixed input... more This paper suggests an improved method for
predictive control of hybrid systems with mixed inputs. The
algorithm takes into account the real nonlinear system as a model
of a hybrid system, which is based on building a tree of evolution.
Where the branch & bound (B&B) technique is applied for
discrete controls in which an embedded nonlinear programming
approach (Pattern search) is associated with each node of the tree
in order to provide the continuous controls and explore the tree.
Once the whole nodes of the tree are explored, the corresponding
input is exploited to the system and the procedure is repeated. The
performance of the resulting predictive control system is
demonstrated on a motorboat simulation case study.
This paper extends the sequential growing and pruning radial basis function (GAP-RBF) to cater fo... more This paper extends the sequential growing and pruning radial basis function (GAP-RBF) to cater for on-line identification of non-linear systems. Some desired modifications on the growing and pruning neurons criteria have been proposed for the original GAP-RBF to make it more suitable for on-line identification. The unscented kalman filter (UKF) has been proposed as a new learning algorithm for GAP-RBF neural network. Moreover, a variable forgetting factor strategy has been included in the UKF algorithm to keep the parameter estimation routine more active in time-varying dynamics. Simulation results show the superiority of the modified GAP-RBF (MGAP-RBF) neural network being learned with the UKF algorithm.
Publisher final version (usually the publisher pdf) Link to publication from Aalborg University C... more Publisher final version (usually the publisher pdf) Link to publication from Aalborg University Citation for published version (APA): Alizadeh, T. (2008). On-line identification of hybrid systems using an adaptive growing and pruning RBF neural
with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of th... more with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of the resulting identification approach is tested and evaluated on a simulated benchmark hybrid system.
Atlas home || Conferences | Abstracts | about Atlas Mathematical Problems in Engineering, Aerospa... more Atlas home || Conferences | Abstracts | about Atlas Mathematical Problems in Engineering, Aerospace and Sciences June 25-27, 2008 University of Genoa, Italy Genoa, Italy. Organizers General Organizer and Chair: Seenith Sivasundaram, USA; Local organizer and Chair: Marcello Sanguineti, Italy. Conference Homepage. Abstracts. Rakhimov Abdumalik Summability of spectral expansions of distributions Lidia Aceto On the relations between block-BVMs and Runge-Kutta collocation methods. ...
This paper suggests an improved method for predictive control of hybrid systems with mixed inputs... more This paper suggests an improved method for predictive control of hybrid systems with mixed inputs. The algorithm takes into account the real nonlinear system as a model of a hybrid system, which is based on building a tree of evolution. Where the branch & bound (B&B) technique is applied for discrete controls in which an embedded nonlinear programming approach (Pattern search) is associated with each node of the tree in order to provide the continuous controls and explore the tree. Once the whole nodes of the tree are explored, the corresponding input is exploited to the system and the procedure is repeated. The performance of the resulting predictive control system is demonstrated on a motorboat simulation case study.
The Matlab/Octave programs on this page are free for academic use. Most of the source codes are i... more The Matlab/Octave programs on this page are free for academic use. Most of the source codes are implementations of the algorithms described in the book "Robot Programming by Demonstration: A Probabilistic Approach". ... Please provide references to the authors' publications in works using parts of these codes. ... The GMM-GMR sourcecode is also available in C/C++ (platform independent) at Sourceforge.net. ... Unzip the file and run 'demo1' in Matlab. ... Calinon, S., Pistillo, A. and Caldwell, DG (2011) Encoding the time and space constraints ...
In this paper we compare two recently proposed algorithms for online identification of hybrid sys... more In this paper we compare two recently proposed algorithms for online identification of hybrid systems. We consider the adaptive growing and pruning Radial Basis Function (RBF) neural network based and the potential fuzzy clustering based procedures. Specific behaviors of the procedures are pointed out, using a well known two dimensional example.
In this paper an innovative method for 1 verification of signature using parametric features base... more In this paper an innovative method for 1 verification of signature using parametric features based on optimal threshold selection is proposed. For each signature, 62 parametric feature are derived from horizontal place, x(t), vertical place, y(t) and pen down and up signals which are obtained from a digitizer plane. The weighted distance between each feature of a signatories and the related reference features is compared to a suitable threshold value and then the feature is accepted or not. The number of the accepted features for a person is then compared to another threshold, which has a suitable value for each signature, and then the signature will be verified or rejected. In this research, 1500 original signatures from 30 person and 600 forgery signatures are used. For each person, 30 genuine and 10 forgery signatures are considered for training of the algorithm and the rest are used in testing and validation. It is shown in the results that there is 0.67% false rejection ratio and 0.67% false acceptation ratio for the training set and a 2.68% and 1.99% for the testing set, respectively.
"Robot learning from demonstrations requires the
robot to learn and adapt movements to new situa... more "Robot learning from demonstrations requires the
robot to learn and adapt movements to new situations, often
characterized by position and orientation of objects or landmarks in the robot’s environment. In the task-parameterized
Gaussian mixture model framework, the movements are considered to be modulated with respect to a set of candidate
frames of reference (coordinate systems) attached to a set of
objects in the robot workspace. Following a similar approach,
this paper addresses the problem of having missing candidate
frames during the demonstrations and reproductions, which
can happen in various situations such as visual occlusion, sensor
unavailability, or tasks with a variable number of descriptive
features. We study this problem with a dust sweeping task in
which the robot requires to consider a variable amount of dust
areas to clean for each reproduction trial."
In contrast to standard industrial robots designed to work in predefined factory settings, humano... more In contrast to standard industrial robots designed to work in predefined factory settings, humanoids are designed to share our space without modifying our environment. These robots unveil new collaboration and social interaction perspectives, but these new environments also expose the robots to various sources of perturbation and unpredictable situations. Since the range of tasks that the humanoids can carry is infinite, it is not possible to provide a predefined database of tasks. For a widespread use in homes and offices, a key requirement will be to provide user-friendly ways of programming the robot to let it acquire new skills and adapt existing skills to new situations. Robot learning by imitation offers a promising route to transfer and refine skills from demonstration by non-expert users.
Gestures are characterized by intermediary or
final landmarks (real or virtual) in task space or... more Gestures are characterized by intermediary or
final landmarks (real or virtual) in task space or joint space
that can change during the course of the motion, and that
are described by varying accuracy and correlation constraints.
Generalizing these trajectories in robot learning by imitation
is challenging, because of the small number of demonstrations
provided by the user. We present an approach to statistically
encode movements in a task-parameterized mixture model,
and derive an expectation-maximization (EM) algorithm to
train it. The model automatically extracts the relevance of
candidate coordinate systems during the task, and exploits this
information during reproduction to adapt the movement in
real-time to changing position and orientation of landmarks or
objects. The approach is tested with a robotic arm learning to
roll out a pizza dough. It is compared to three categories of taskparameterized
models: 1) Gaussian process regression (GPR)
with a trajectory models database; 2) Multi-streams approach
with models trained in several frames of reference; and 3)
Parametric Gaussian mixture model (PGMM) modulating the
Gaussian centers with the task parameters. We show that the
extrapolation capability of the proposed approach outperforms
existing methods, by extracting the local structures of the task
instead of relying on interpolation principles.
with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of th... more with forgetting factor scheme as the required on-line learning algorithm. The effectiveness of the resulting identification approach is tested and evaluated on a simulated benchmark hybrid system.
This paper suggests an improved method for
predictive control of hybrid systems with mixed input... more This paper suggests an improved method for
predictive control of hybrid systems with mixed inputs. The
algorithm takes into account the real nonlinear system as a model
of a hybrid system, which is based on building a tree of evolution.
Where the branch & bound (B&B) technique is applied for
discrete controls in which an embedded nonlinear programming
approach (Pattern search) is associated with each node of the tree
in order to provide the continuous controls and explore the tree.
Once the whole nodes of the tree are explored, the corresponding
input is exploited to the system and the procedure is repeated. The
performance of the resulting predictive control system is
demonstrated on a motorboat simulation case study.
This paper extends the sequential growing and pruning radial basis function (GAP-RBF) to cater fo... more This paper extends the sequential growing and pruning radial basis function (GAP-RBF) to cater for on-line identification of non-linear systems. Some desired modifications on the growing and pruning neurons criteria have been proposed for the original GAP-RBF to make it more suitable for on-line identification. The unscented kalman filter (UKF) has been proposed as a new learning algorithm for GAP-RBF neural network. Moreover, a variable forgetting factor strategy has been included in the UKF algorithm to keep the parameter estimation routine more active in time-varying dynamics. Simulation results show the superiority of the modified GAP-RBF (MGAP-RBF) neural network being learned with the UKF algorithm.
Publisher final version (usually the publisher pdf) Link to publication from Aalborg University C... more Publisher final version (usually the publisher pdf) Link to publication from Aalborg University Citation for published version (APA): Alizadeh, T. (2008). On-line identification of hybrid systems using an adaptive growing and pruning RBF neural
Uploads
Papers by Tohid Alizadeh
robot to learn and adapt movements to new situations, often
characterized by position and orientation of objects or landmarks in the robot’s environment. In the task-parameterized
Gaussian mixture model framework, the movements are considered to be modulated with respect to a set of candidate
frames of reference (coordinate systems) attached to a set of
objects in the robot workspace. Following a similar approach,
this paper addresses the problem of having missing candidate
frames during the demonstrations and reproductions, which
can happen in various situations such as visual occlusion, sensor
unavailability, or tasks with a variable number of descriptive
features. We study this problem with a dust sweeping task in
which the robot requires to consider a variable amount of dust
areas to clean for each reproduction trial."
final landmarks (real or virtual) in task space or joint space
that can change during the course of the motion, and that
are described by varying accuracy and correlation constraints.
Generalizing these trajectories in robot learning by imitation
is challenging, because of the small number of demonstrations
provided by the user. We present an approach to statistically
encode movements in a task-parameterized mixture model,
and derive an expectation-maximization (EM) algorithm to
train it. The model automatically extracts the relevance of
candidate coordinate systems during the task, and exploits this
information during reproduction to adapt the movement in
real-time to changing position and orientation of landmarks or
objects. The approach is tested with a robotic arm learning to
roll out a pizza dough. It is compared to three categories of taskparameterized
models: 1) Gaussian process regression (GPR)
with a trajectory models database; 2) Multi-streams approach
with models trained in several frames of reference; and 3)
Parametric Gaussian mixture model (PGMM) modulating the
Gaussian centers with the task parameters. We show that the
extrapolation capability of the proposed approach outperforms
existing methods, by extracting the local structures of the task
instead of relying on interpolation principles.
predictive control of hybrid systems with mixed inputs. The
algorithm takes into account the real nonlinear system as a model
of a hybrid system, which is based on building a tree of evolution.
Where the branch & bound (B&B) technique is applied for
discrete controls in which an embedded nonlinear programming
approach (Pattern search) is associated with each node of the tree
in order to provide the continuous controls and explore the tree.
Once the whole nodes of the tree are explored, the corresponding
input is exploited to the system and the procedure is repeated. The
performance of the resulting predictive control system is
demonstrated on a motorboat simulation case study.
robot to learn and adapt movements to new situations, often
characterized by position and orientation of objects or landmarks in the robot’s environment. In the task-parameterized
Gaussian mixture model framework, the movements are considered to be modulated with respect to a set of candidate
frames of reference (coordinate systems) attached to a set of
objects in the robot workspace. Following a similar approach,
this paper addresses the problem of having missing candidate
frames during the demonstrations and reproductions, which
can happen in various situations such as visual occlusion, sensor
unavailability, or tasks with a variable number of descriptive
features. We study this problem with a dust sweeping task in
which the robot requires to consider a variable amount of dust
areas to clean for each reproduction trial."
final landmarks (real or virtual) in task space or joint space
that can change during the course of the motion, and that
are described by varying accuracy and correlation constraints.
Generalizing these trajectories in robot learning by imitation
is challenging, because of the small number of demonstrations
provided by the user. We present an approach to statistically
encode movements in a task-parameterized mixture model,
and derive an expectation-maximization (EM) algorithm to
train it. The model automatically extracts the relevance of
candidate coordinate systems during the task, and exploits this
information during reproduction to adapt the movement in
real-time to changing position and orientation of landmarks or
objects. The approach is tested with a robotic arm learning to
roll out a pizza dough. It is compared to three categories of taskparameterized
models: 1) Gaussian process regression (GPR)
with a trajectory models database; 2) Multi-streams approach
with models trained in several frames of reference; and 3)
Parametric Gaussian mixture model (PGMM) modulating the
Gaussian centers with the task parameters. We show that the
extrapolation capability of the proposed approach outperforms
existing methods, by extracting the local structures of the task
instead of relying on interpolation principles.
predictive control of hybrid systems with mixed inputs. The
algorithm takes into account the real nonlinear system as a model
of a hybrid system, which is based on building a tree of evolution.
Where the branch & bound (B&B) technique is applied for
discrete controls in which an embedded nonlinear programming
approach (Pattern search) is associated with each node of the tree
in order to provide the continuous controls and explore the tree.
Once the whole nodes of the tree are explored, the corresponding
input is exploited to the system and the procedure is repeated. The
performance of the resulting predictive control system is
demonstrated on a motorboat simulation case study.