Controlling hybrid systems - a system that exhibits continuous and discrete behavior simultaneous... more Controlling hybrid systems - a system that exhibits continuous and discrete behavior simultaneously - is of great interest since the new millennium. Switched linear systems are especially interesting due to the large amount of applications that may be solved. However, applying different control schemes on switched systems entails difficulties in identifying the underlying models and the transitions that occur between them. In this paper an automatic identification and switching for Multi-Model Reference Adaptive Control (MMRAC) scheme is proposed. The identification of the submodels is performed by curve clustering of the states plotted in the phase portrait. An unsupervised learning algorithm is proposed to cluster the curves. Each curve represents a single submodel and is paired with an MRAC. After the clustering process, correlation between every submodel and the current state is checked. Then the MRAC paired with the best representing curve is used to control the plant, and update the parameters of the curve and the MRAC itself. The results of two simulations are presented in the end of this paper.
In this paper we present a novel mathematical model for color image processing. The proposed alge... more In this paper we present a novel mathematical model for color image processing. The proposed algebraic structure is based on a special mapping of color vectors into the space of bi-quaternions (quaternions with complex coefficients) inspired by the theory of relativity. By this transformation, the space of color vectors remains close under scalar multiplication and addition and limited by an upper and lower bounds. The proposed approach is therefore terms Cages Image Processing (KIP). We demonstrate the usability of the new model by a color image enhancement algorithm. The proposed enhancement algorithm prevents information loss caused by color saturation which is an artifact of the Logarithmic Image Processing (LIP) approach. Experimental results on synthetic and natural images comparing the proposed algorithm to the LIP based algorithm are provided.
This paper presents a solution to the trajectory-tracking control problem of an Autonomous Surfac... more This paper presents a solution to the trajectory-tracking control problem of an Autonomous Surface Vehicle (ASV) under the effect of sea waves. Worldwide, there has been increasing interest in the use of ASVs to execute missions of increasing complexity without direct supervision of human operators. A key-enabling element for the execution of such missions is the availability of advanced systems for motion control of ASVs. The problems of motion control can be roughly classified into three groups: i) point stabilization, where the goal is to stabilize a vehicle at a given target point with a desired orientation; ii) trajectory tracking, where the vehicle is required to track a time parameterized reference; and iii) path following, where the objective is for the vehicle to converge to and follow a desired geometric path, without an explicit timing law assigned to it. In some applications where it is critical for ASVs to follow the preplanning trajectory as closely as possible, the trajectory problem of ASVs becomes important.
In the last decade many studies have been conducted in the technological framework of Unmanned Ae... more In the last decade many studies have been conducted in the technological framework of Unmanned Aerial Vehicles (UAVs). There is great interest in developing autonomous UAVs, in particular, unmanned vertical take-off and landing (UVTOL) aircraft, because of their ability to conduct search and rescue missions in remote dangerous areas, without endangering human operators. The highly complicated flight control system of UVTOL aircraft presents a serious challenge for scientists. In this research we developed a control algorithm for quadrotor attitude and altitude stabilization. The system, which includes advanced hardware for real-time data processing, is reliable and robust and thus ensures safe flight and take-off and landing. In the current study we implement the control scheme in an analytical model of a quadrotor platform and demonstrate vertical take-off and landing and hovering. Later, the realization of the proposed controller will be based on microsensors and microelectronic c...
Medial axis transform (MAT) based features and a multilayer perceptron (MLP) neural network (NN) ... more Medial axis transform (MAT) based features and a multilayer perceptron (MLP) neural network (NN) were used for human chromosome classification. Two approaches to the MAT, one based on skeletonization and the other based on a piecewise linear (PWL) ...
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016
Controlled light conditions improve considerably the performance of most computer vision algorith... more Controlled light conditions improve considerably the performance of most computer vision algorithms. Dynamic light conditions create varying spatial changes in color and intensity across the scene. These condition, caused by a moving shadow for example, force developers to create algorithms which are robust to such variations. We suggest a computational camera which produces images that are not influenced by environmental variations in light conditions. The key insight is that many years ago, similar difficulties were already solved in radio communication, As a result each channel is immune to interference from other radio channels. Amplitude Modulated (AM) video camera separates the influence of a modulated light from other unknown light sources in the scene, Causing the AM video camera frame to appear the same - independent of the light conditions in which it was taken. We built a prototype of the AM video camera by using off the shelf hardware and tested it. AM video camera was u...
: The objective of this research project was to develop and provide a dedicated man-machine inter... more : The objective of this research project was to develop and provide a dedicated man-machine interaction method tailored for supervised autonomous systems. Implementation of an optimal methodology for human-robot cooperation involved combining the advantages of human perception and recognition skills with the autonomous robots accuracy and consistency. A cooperative human robotic system can increase identification capabilities, and enable recognition even in unpredictable conditions that autonomous systems are incompetent to deal with. Providing means for optional, high-level human intervention, along with pre-acquired machine skills, stands at the base of our supervised autonomy control concept. The expected impact is to enable smooth real-time adaptation of the combined human robot system to many possible changes of the environment and performances of both robot and human operator.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE)
For an intelligent autonomous underwater system to interact with its environment, it needs to ach... more For an intelligent autonomous underwater system to interact with its environment, it needs to achieve a navigation solution as accurate as possible. Traditional dead reckoning approach can only achieve a low drift error; therefore, the position accuracy monotonically decreases with distance traveled. Approaches such as simultaneous localization and mapping (SLAM) and sharing location information between different vehicles can improve localization accuracy. The approaches for Autonomous Underwater Vehicle (AUV) navigation and preliminary experimental and simulations results will be presented in this work.
Speaker diarization systems attempt to assign temporal speech segments in a conversation to the a... more Speaker diarization systems attempt to assign temporal speech segments in a conversation to the appropriate speaker, and non-speech segments to non-speech. Speaker diarization systems basically provide an answer to the question "Who spoke when ?". One inherent deficiency of most current systems is their inability to handle co-channel or overlapped speech. During the past few years, several studies have attempted dealing with the problem of overlapped or co-channel speech detection and separation, however, most of the algorithms suggested perform under unique conditions, require high computational complexity and require both time and frequency domain analysis of the audio data. In this study, frame based entropy analysis of the audio data in the time domain serves as a single feature for an overlapped speech detection algorithm. Identification of overlapped speech segments is performed using Gaussian Mixture Modeling (GMM) along with well known classification algorithms app...
The underwater environment makes object-detection missions difficult. Side-scan Sonar (SSS) has b... more The underwater environment makes object-detection missions difficult. Side-scan Sonar (SSS) has been found to be suitable for seabed scanning missions, however the sonar images acquired from SSS often suffer from considerable noise and geometrical distortion, which changes the understanding of the texture, size, and shape of seabed objects. In order to identify seabed objects, it is thus vital to reconstruct the actual shape by reducing distortion. This paper proposes a process for correcting and reconstructing the sonar image map that utilizes intensity normalization, slant range correction, yaw and pitch correction, and speed and location correction. This is done using navigation and inertial data acquired by the autonomous underwater vehicle sensors.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE), 2016
Object appearance can change dramatically from one image to another. One of the main reasons for ... more Object appearance can change dramatically from one image to another. One of the main reasons for this is light variations. We created a video camera that is invariant to light conditions. Light condition can change in color, orientation or pattern. The system is based on principles from radio communication which is able to separates one station from the other. In a similar manner we modulate light intensity in time, and its influence on the scene is separated from other light sources using a match filter. The modulated light source is completely independent and is not synchronized with the camera. This simplifies the system considerably and enables shadow free objects with constant color. Therefore objects will always appear the same under any light conditions. This has the potential to improve performance of many machine vision algorithms. We demonstrate this by showing a performance improvement on commercially available face recognition system.
Underwater navigation is a challenging task for Autonomous Underwater Vehicles (AUVs), which requ... more Underwater navigation is a challenging task for Autonomous Underwater Vehicles (AUVs), which requires to estimate the position and orientation of the vehicle. Accelerometer-based attitude estimation is problematic as its accuracy can degrade by up to two orders of magnitude when the AUV is accelerating. Pressure sensors, which are unaffected by acceleration, offer an efficient alternative for determining orientation. This paper offers a method for pitch-and-roll estimation using a network of pressure sensors and a thorough analysis of the method’s accuracy including analytical evaluation and the Cramér-Rao lower bound. Then, the method is tested in a simulation. We also provide a way to estimate system performance based on the vehicle’s size and the number and accuracy of the sensors.
Autonomous vehicles (AVs) are expected to be on public roads in the near future, but some critica... more Autonomous vehicles (AVs) are expected to be on public roads in the near future, but some critical aspects of this reality are yet to be resolved. One of these aspects is the lack of a sufficient safety-performance-verification technique. The existing tools of functional-safety engineering do not provide a comprehensive solution for the verification of artificial intelligence and machine learning. For the verification of this type of algorithms, it has become widely accepted in the recent years that simulation tests should be used. Nevertheless, to the best of our knowledge, no detailed method for how these simulation tests should be performed has yet been suggested. To start tackling this gap, this paper presents a verification methodology based on the statistical testing of AVs’ safety-related functionality in simulated scenarios. Also, presented here is a test-case implementation of the methodology on a full-scale autonomous unmanned ground vehicle (UGV). In this example, the functionality of safe autonomous off-road navigation is verified. It is shown that the statistical performance data that are gathered using simulation can be used to accurately predict the overall safety performance of UGVs in real life. Our intention is that this paper will provide a starting point for the further discussion on the exact way simulation tests shall be used for the safety verification of AVs.
Human ESCs (hESCs) are a valuable tool for the study of early human development and represent a s... more Human ESCs (hESCs) are a valuable tool for the study of early human development and represent a source of normal differentiated cells for pharmaceutical and biotechnology applications and ultimately for cell replacement therapies. For all applications, it will be necessary to develop assays to validate the efficacy of hESC differentiation. We explored the capacity for FTIR spectroscopy, a technique that
IEEE Transactions on Instrumentation and Measurement, 1990
Various digital signal processing methods, which could be applicable to the design of a microcomp... more Various digital signal processing methods, which could be applicable to the design of a microcomputer-based thermal pulse flowmeter, are examined. Two excitation methods are investigated: a single thermal pulse, and a pseudorandom binary sequence signal (PRBS). The signal recovered downstream is processed by two alternative numerical algorithms to recover the time of flight (i.e. by peak detection of the signal
ABSTRACT An auto-adaptive neuro-fuzzy segmentation architecture is presented. The system consists... more ABSTRACT An auto-adaptive neuro-fuzzy segmentation architecture is presented. The system consists of a multilayer perceptron (MLP) network that performs adaptive thresholding of the input image using labels automatically preselected by a fuzzy clustering technique. The proposed architecture is feedforward, but unlike the conventional MLP the learning is unsupervised. The output status of the network is described as a fuzzy set. Fuzzy entropy is used as a measure of the error of the system
Controlling hybrid systems - a system that exhibits continuous and discrete behavior simultaneous... more Controlling hybrid systems - a system that exhibits continuous and discrete behavior simultaneously - is of great interest since the new millennium. Switched linear systems are especially interesting due to the large amount of applications that may be solved. However, applying different control schemes on switched systems entails difficulties in identifying the underlying models and the transitions that occur between them. In this paper an automatic identification and switching for Multi-Model Reference Adaptive Control (MMRAC) scheme is proposed. The identification of the submodels is performed by curve clustering of the states plotted in the phase portrait. An unsupervised learning algorithm is proposed to cluster the curves. Each curve represents a single submodel and is paired with an MRAC. After the clustering process, correlation between every submodel and the current state is checked. Then the MRAC paired with the best representing curve is used to control the plant, and update the parameters of the curve and the MRAC itself. The results of two simulations are presented in the end of this paper.
In this paper we present a novel mathematical model for color image processing. The proposed alge... more In this paper we present a novel mathematical model for color image processing. The proposed algebraic structure is based on a special mapping of color vectors into the space of bi-quaternions (quaternions with complex coefficients) inspired by the theory of relativity. By this transformation, the space of color vectors remains close under scalar multiplication and addition and limited by an upper and lower bounds. The proposed approach is therefore terms Cages Image Processing (KIP). We demonstrate the usability of the new model by a color image enhancement algorithm. The proposed enhancement algorithm prevents information loss caused by color saturation which is an artifact of the Logarithmic Image Processing (LIP) approach. Experimental results on synthetic and natural images comparing the proposed algorithm to the LIP based algorithm are provided.
This paper presents a solution to the trajectory-tracking control problem of an Autonomous Surfac... more This paper presents a solution to the trajectory-tracking control problem of an Autonomous Surface Vehicle (ASV) under the effect of sea waves. Worldwide, there has been increasing interest in the use of ASVs to execute missions of increasing complexity without direct supervision of human operators. A key-enabling element for the execution of such missions is the availability of advanced systems for motion control of ASVs. The problems of motion control can be roughly classified into three groups: i) point stabilization, where the goal is to stabilize a vehicle at a given target point with a desired orientation; ii) trajectory tracking, where the vehicle is required to track a time parameterized reference; and iii) path following, where the objective is for the vehicle to converge to and follow a desired geometric path, without an explicit timing law assigned to it. In some applications where it is critical for ASVs to follow the preplanning trajectory as closely as possible, the trajectory problem of ASVs becomes important.
In the last decade many studies have been conducted in the technological framework of Unmanned Ae... more In the last decade many studies have been conducted in the technological framework of Unmanned Aerial Vehicles (UAVs). There is great interest in developing autonomous UAVs, in particular, unmanned vertical take-off and landing (UVTOL) aircraft, because of their ability to conduct search and rescue missions in remote dangerous areas, without endangering human operators. The highly complicated flight control system of UVTOL aircraft presents a serious challenge for scientists. In this research we developed a control algorithm for quadrotor attitude and altitude stabilization. The system, which includes advanced hardware for real-time data processing, is reliable and robust and thus ensures safe flight and take-off and landing. In the current study we implement the control scheme in an analytical model of a quadrotor platform and demonstrate vertical take-off and landing and hovering. Later, the realization of the proposed controller will be based on microsensors and microelectronic c...
Medial axis transform (MAT) based features and a multilayer perceptron (MLP) neural network (NN) ... more Medial axis transform (MAT) based features and a multilayer perceptron (MLP) neural network (NN) were used for human chromosome classification. Two approaches to the MAT, one based on skeletonization and the other based on a piecewise linear (PWL) ...
2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016
Controlled light conditions improve considerably the performance of most computer vision algorith... more Controlled light conditions improve considerably the performance of most computer vision algorithms. Dynamic light conditions create varying spatial changes in color and intensity across the scene. These condition, caused by a moving shadow for example, force developers to create algorithms which are robust to such variations. We suggest a computational camera which produces images that are not influenced by environmental variations in light conditions. The key insight is that many years ago, similar difficulties were already solved in radio communication, As a result each channel is immune to interference from other radio channels. Amplitude Modulated (AM) video camera separates the influence of a modulated light from other unknown light sources in the scene, Causing the AM video camera frame to appear the same - independent of the light conditions in which it was taken. We built a prototype of the AM video camera by using off the shelf hardware and tested it. AM video camera was u...
: The objective of this research project was to develop and provide a dedicated man-machine inter... more : The objective of this research project was to develop and provide a dedicated man-machine interaction method tailored for supervised autonomous systems. Implementation of an optimal methodology for human-robot cooperation involved combining the advantages of human perception and recognition skills with the autonomous robots accuracy and consistency. A cooperative human robotic system can increase identification capabilities, and enable recognition even in unpredictable conditions that autonomous systems are incompetent to deal with. Providing means for optional, high-level human intervention, along with pre-acquired machine skills, stands at the base of our supervised autonomy control concept. The expected impact is to enable smooth real-time adaptation of the combined human robot system to many possible changes of the environment and performances of both robot and human operator.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE)
For an intelligent autonomous underwater system to interact with its environment, it needs to ach... more For an intelligent autonomous underwater system to interact with its environment, it needs to achieve a navigation solution as accurate as possible. Traditional dead reckoning approach can only achieve a low drift error; therefore, the position accuracy monotonically decreases with distance traveled. Approaches such as simultaneous localization and mapping (SLAM) and sharing location information between different vehicles can improve localization accuracy. The approaches for Autonomous Underwater Vehicle (AUV) navigation and preliminary experimental and simulations results will be presented in this work.
Speaker diarization systems attempt to assign temporal speech segments in a conversation to the a... more Speaker diarization systems attempt to assign temporal speech segments in a conversation to the appropriate speaker, and non-speech segments to non-speech. Speaker diarization systems basically provide an answer to the question "Who spoke when ?". One inherent deficiency of most current systems is their inability to handle co-channel or overlapped speech. During the past few years, several studies have attempted dealing with the problem of overlapped or co-channel speech detection and separation, however, most of the algorithms suggested perform under unique conditions, require high computational complexity and require both time and frequency domain analysis of the audio data. In this study, frame based entropy analysis of the audio data in the time domain serves as a single feature for an overlapped speech detection algorithm. Identification of overlapped speech segments is performed using Gaussian Mixture Modeling (GMM) along with well known classification algorithms app...
The underwater environment makes object-detection missions difficult. Side-scan Sonar (SSS) has b... more The underwater environment makes object-detection missions difficult. Side-scan Sonar (SSS) has been found to be suitable for seabed scanning missions, however the sonar images acquired from SSS often suffer from considerable noise and geometrical distortion, which changes the understanding of the texture, size, and shape of seabed objects. In order to identify seabed objects, it is thus vital to reconstruct the actual shape by reducing distortion. This paper proposes a process for correcting and reconstructing the sonar image map that utilizes intensity normalization, slant range correction, yaw and pitch correction, and speed and location correction. This is done using navigation and inertial data acquired by the autonomous underwater vehicle sensors.
2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE), 2016
Object appearance can change dramatically from one image to another. One of the main reasons for ... more Object appearance can change dramatically from one image to another. One of the main reasons for this is light variations. We created a video camera that is invariant to light conditions. Light condition can change in color, orientation or pattern. The system is based on principles from radio communication which is able to separates one station from the other. In a similar manner we modulate light intensity in time, and its influence on the scene is separated from other light sources using a match filter. The modulated light source is completely independent and is not synchronized with the camera. This simplifies the system considerably and enables shadow free objects with constant color. Therefore objects will always appear the same under any light conditions. This has the potential to improve performance of many machine vision algorithms. We demonstrate this by showing a performance improvement on commercially available face recognition system.
Underwater navigation is a challenging task for Autonomous Underwater Vehicles (AUVs), which requ... more Underwater navigation is a challenging task for Autonomous Underwater Vehicles (AUVs), which requires to estimate the position and orientation of the vehicle. Accelerometer-based attitude estimation is problematic as its accuracy can degrade by up to two orders of magnitude when the AUV is accelerating. Pressure sensors, which are unaffected by acceleration, offer an efficient alternative for determining orientation. This paper offers a method for pitch-and-roll estimation using a network of pressure sensors and a thorough analysis of the method’s accuracy including analytical evaluation and the Cramér-Rao lower bound. Then, the method is tested in a simulation. We also provide a way to estimate system performance based on the vehicle’s size and the number and accuracy of the sensors.
Autonomous vehicles (AVs) are expected to be on public roads in the near future, but some critica... more Autonomous vehicles (AVs) are expected to be on public roads in the near future, but some critical aspects of this reality are yet to be resolved. One of these aspects is the lack of a sufficient safety-performance-verification technique. The existing tools of functional-safety engineering do not provide a comprehensive solution for the verification of artificial intelligence and machine learning. For the verification of this type of algorithms, it has become widely accepted in the recent years that simulation tests should be used. Nevertheless, to the best of our knowledge, no detailed method for how these simulation tests should be performed has yet been suggested. To start tackling this gap, this paper presents a verification methodology based on the statistical testing of AVs’ safety-related functionality in simulated scenarios. Also, presented here is a test-case implementation of the methodology on a full-scale autonomous unmanned ground vehicle (UGV). In this example, the functionality of safe autonomous off-road navigation is verified. It is shown that the statistical performance data that are gathered using simulation can be used to accurately predict the overall safety performance of UGVs in real life. Our intention is that this paper will provide a starting point for the further discussion on the exact way simulation tests shall be used for the safety verification of AVs.
Human ESCs (hESCs) are a valuable tool for the study of early human development and represent a s... more Human ESCs (hESCs) are a valuable tool for the study of early human development and represent a source of normal differentiated cells for pharmaceutical and biotechnology applications and ultimately for cell replacement therapies. For all applications, it will be necessary to develop assays to validate the efficacy of hESC differentiation. We explored the capacity for FTIR spectroscopy, a technique that
IEEE Transactions on Instrumentation and Measurement, 1990
Various digital signal processing methods, which could be applicable to the design of a microcomp... more Various digital signal processing methods, which could be applicable to the design of a microcomputer-based thermal pulse flowmeter, are examined. Two excitation methods are investigated: a single thermal pulse, and a pseudorandom binary sequence signal (PRBS). The signal recovered downstream is processed by two alternative numerical algorithms to recover the time of flight (i.e. by peak detection of the signal
ABSTRACT An auto-adaptive neuro-fuzzy segmentation architecture is presented. The system consists... more ABSTRACT An auto-adaptive neuro-fuzzy segmentation architecture is presented. The system consists of a multilayer perceptron (MLP) network that performs adaptive thresholding of the input image using labels automatically preselected by a fuzzy clustering technique. The proposed architecture is feedforward, but unlike the conventional MLP the learning is unsupervised. The output status of the network is described as a fuzzy set. Fuzzy entropy is used as a measure of the error of the system
Uploads
Papers by Hugo Guterman