Abstract Following surgical repair of major nerve trunks there is a rapid functional reorganisation in the brain, characterised by a changed cortical mapping of the hand secondary to incorrect reinnervation of peripheral targets. Sensory... more
Abstract Following surgical repair of major nerve trunks there is a rapid functional reorganisation in the brain, characterised by a changed cortical mapping of the hand secondary to incorrect reinnervation of peripheral targets. Sensory re-educational ...
Research Interests:
A reinforcement architecture is introduced that consists of three complementary learning systems with different generalization abilities. The ACTOR learns state-action associations, the CRITIC learns a goal-gradient, and the PUNISH system... more
A reinforcement architecture is introduced that consists of three complementary learning systems with different generalization abilities. The ACTOR learns state-action associations, the CRITIC learns a goal-gradient, and the PUNISH system learns what actions to avoid. The architecture is compared to the standard actor-crititc and Q-learning models on a number of maze learning tasks. The novel architecture is shown to be superior on all the test mazes. Moreover, it shows how it is possible to combine several learning systems with different properties in a coherent reinforcement learning framework.
The core deficit in attention deficit/hyperactivity disorder (ADHD) continues to present an enigma to scientists, professionals and parents alike. Medication with methylphenidate has to some extent proven successful, but it is still not... more
The core deficit in attention deficit/hyperactivity disorder (ADHD) continues to present an enigma to scientists, professionals and parents alike. Medication with methylphenidate has to some extent proven successful, but it is still not known exactly what parts of the deficits in ADHD medication influences. It is thus important to further delineate the nature of ADHD and thoroughly investigate the possible subtypes. In this paper we describe some experiments investigating the nature of inhibition, resource allocation, sustained and selective attention, distractibility and task switching among children with ADHD. The conclusions could be used to advocate for different models of ADHD, e.g., a model of deficient response inhibition or a model of non-optimal resource allocation. Below we present our first steps toward a robot model of ADHD. The computational model can reproduce some of the behaviors shown in children with ADHD in experimental situations. The model suggests that a defici...
Research Interests:
Introduction To act in an unknown and continously changing environment, an autonomous robot must be able to react instantaneously on changes and unexpected events in order to avoid collisions and to update its maps. Successful navigation... more
Introduction To act in an unknown and continously changing environment, an autonomous robot must be able to react instantaneously on changes and unexpected events in order to avoid collisions and to update its maps. Successful navigation requires that the robot reacts primarily on its immediate sensory information and secondarily on its internal mapping of the spatial layout of the environment. We have developed and constructed an experimental mobile robot equipped with a number of complementary sensory systems (Balkenius and Kopp 1994a). A video camera is mounted on a movable head that also contains a pair of microphones. Ultrasonic sensors are located around the body of the robot and a set of tactile sensors (whiskers) and a bumper are used to detect obstacles at a short range. The project aims at developing the attention and navigation systems of the robot to include vision for spatial orientation. The choice of vision is natural since this modality contains the richest informati
No current commercially available myoelectrically controlled prosthetic hands provide conscious sensory feedback to the user. A system aiming at relocation of sensory input from a prosthetic hand equipped with force sensors to the forearm... more
No current commercially available myoelectrically controlled prosthetic hands provide conscious sensory feedback to the user. A system aiming at relocation of sensory input from a prosthetic hand equipped with force sensors to the forearm skin of amputees, a tactile display, has been developed and constructed. The system consists of five piezoresistive force sensors or, alternatively, a prosthetic hand equipped with force sensors, five digital servomotors with a lever and a circular plastic disk pushing on the skin, control electronics based on an MSP430 microcontroller and a test application implemented in LabVIEW running on a PC. The tactile display system is intended to be integrated into the socket of a hand prosthesis and used as a conscious sensory feedback system for hand amputees using a myoelectrically controlled hand prosthesis. The system will provide continuous force feedback from sensors in the fingertips of each prosthetic finger and will likely improve the users' ...
BERRY III is an interactive tool for the construction and simulation of neural network controlled artificial creatures that can be considered either as biological models or as simulated robots. Environments and creatures can be built and... more
BERRY III is an interactive tool for the construction and simulation of neural network controlled artificial creatures that can be considered either as biological models or as simulated robots. Environments and creatures can be built and modified during simulation time using a large set of graphical editors and viewers. The simulator also includes an extensive package for generation of graphs and statistics. To let the neural network controllers improve automatically, it possible to invoke an evolution module that uses the genetic algorithm to create new neural networks using a large selection of evolution schemes. The simulator is in the public domain and suitable both as a research tool and in education.
We describe how a standard reinforcement learning algorithm can be changed to include a second contextual input that is used to modulate the learning in the original algorithm. The new algorithm takes the context into account during... more
We describe how a standard reinforcement learning algorithm can be changed to include a second contextual input that is used to modulate the learning in the original algorithm. The new algorithm takes the context into account during relearning when the previously learned actions are no longer valid. The algorithm was tested on a number of cognitive experiment and shown to reproduce the learning in both a task switching test and in the Wisconsin Card Sorting Test. In addition, the algorithm was...
Research Interests:
The Artificial Hand Project is a cross disciplinary cooperation project group that has the vision of building an artificial hand that will be accepted by the bearer as a part of the body and not as a tool as it is today. The investigation... more
The Artificial Hand Project is a cross disciplinary cooperation project group that has the vision of building an artificial hand that will be accepted by the bearer as a part of the body and not as a tool as it is today. The investigation of a control system for such an advanced prosthesis is carried out in this paper using
Research Interests:
By using a microfabricated nerve chip with integrated electrodes through which peripheral nerves can regenerate and make electrical connects, it should be possible to control a remote prosthesis by processing the detected nerve signals.... more
By using a microfabricated nerve chip with integrated electrodes through which peripheral nerves can regenerate and make electrical connects, it should be possible to control a remote prosthesis by processing the detected nerve signals. In this study different artificial neural networks have been employed for classification of such complex patterns of signals. The signals were obtained from four electrodes detecting muscle activity in a rat hindlimb as a consequence of applied stimulus to the rat right hindpaw. These signals recorded at four different sites resembles a situation of a nerve chip with four electrodes, which implies that one might be able to use the same strategy when analyzing data from a four-electrode chip to obtain information from the nervous system. Here, the authors address the usefulness of different network topologies for analyzing measured in vivo data from an implanted perforated nerve chip
Research Interests:
The Double Cone Model (DCM) is a model of how the brain transforms sensory input to motor commands through successive stages of data compression and expansion. We have tested a subset of the DCM on speech recog- nition, production and... more
The Double Cone Model (DCM) is a model of how the brain transforms sensory input to motor commands through successive stages of data compression and expansion. We have tested a subset of the DCM on speech recog- nition, production and imitation. The experi- ments show that the DCM is a good candidate for an articial speech processing system that can
Research Interests:
To accurately track a moving target despite delays in the visual processing, it is necessary to predict how the target will move in the future and to direct the gaze toward the anticipated location. We have ad-dressed this problem within... more
To accurately track a moving target despite delays in the visual processing, it is necessary to predict how the target will move in the future and to direct the gaze toward the anticipated location. We have ad-dressed this problem within a simulated visual scene were a ball is dropped from a random location and subsequently bounces or rolls on different objects be-fore it comes to rest. A developmental system has been designed that attempts to track the location of the ball as closely as possible despite a delay of 250 ms from the the visual input to the control of gaze. The system is an extension of an earlier model of gaze development (Balkenius and Johansson, 2005). A standard Kalman filter-based tracking frame-work is used that gradually develops a set of parallel predictive models of the environment. The use of multiple models is similar to the MOSAIC architec-ture (Wolpert et al., 2003). These models are used to predict the future position of a target object, and by generalizi...
This paper presents an experimental study using two robots. In the experiment, the robots navigated through an area with or without obstacles and had as their goal to shift places with each other. Four dif-ferent approaches (random,... more
This paper presents an experimental study using two robots. In the experiment, the robots navigated through an area with or without obstacles and had as their goal to shift places with each other. Four dif-ferent approaches (random, reactive, planning, anticipation) were used during the experiment and the times to accomplish the task were com-pered. The results indicate that the ability to anticipate the behavior of the other robot could be to an advantage. However, the results also clearly show that anticipatory behavior is not always better than a purely reactive strategy.
This report describes a fast method for finding regions with a particular color in a visual scene. The method is general enough to be used in any situations were a number of objects with known color need to be found in an image. We demon-... more
This report describes a fast method for finding regions with a particular color in a visual scene. The method is general enough to be used in any situations were a number of objects with known color need to be found in an image. We demon- strate how the algorithm can be used to find and track robots wi th colored markers, symbols with a particular color as well as for simple color-based face detection.
Research Interests:
A model of event driven anticipatory learn-ing is described and applied to a number of attention situations where one or several vi-sual targets need to be tracked while being intermittently occluded. The model combines covert tracking of... more
A model of event driven anticipatory learn-ing is described and applied to a number of attention situations where one or several vi-sual targets need to be tracked while being intermittently occluded. The model combines covert tracking of multiple targets with overt control of a single attention focus. The im-plemented system has been applied to both a simple scenario with a car that is occluded in a tunnel and a complex situation with six simu-lated robots that need to anticipate the move-ments of each other. The system is shown to learn very quickly to anticipate target move-ments. The performance is further increased when the simulated robots are allowed to co-operate in the tracking task.
We have implemented four bio-inspired self- organizing haptic systems based on proprioception on a 12 d.o.f. anthropomorphic robot hand. The four systems differ in the kind of self-organizing neural network used for clustering. For the... more
We have implemented four bio-inspired self- organizing haptic systems based on proprioception on a 12 d.o.f. anthropomorphic robot hand. The four systems differ in the kind of self-organizing neural network used for clustering. For the mapping of the explored objects, one system uses a Self- Organizing Map (SOM), one uses a Growing Cell Structure (GCS), one uses a Growing Cell
Research Interests:
We have experimented with a bio-inspired self- organizing texture and hardness perception system which automatically learns to associate the representations of the two submodalities with each other. To this end we have developed a... more
We have experimented with a bio-inspired self- organizing texture and hardness perception system which automatically learns to associate the representations of the two submodalities with each other. To this end we have developed a microphone based texture sensor and a hardness sensor that measures the compression of the material at a constant pressure. The system is based on a novel
Research Interests:
The Lucs Haptic Hand III has been built as a step in a project at LUCS aiming at studying haptic perception. In this project, several robot hands together with cog- nitive computational models of the corresponding hu- man neurophysiologic... more
The Lucs Haptic Hand III has been built as a step in a project at LUCS aiming at studying haptic perception. In this project, several robot hands together with cog- nitive computational models of the corresponding hu- man neurophysiologic systems will be built. Grasping tests with the LUCS Haptic Hand III were done with six different objects in order to
Research Interests:
We have developed an 8 d.o.f. robot hand which has been tested with three computational models of haptic perception. Two of the models are based on the tensor product of different proprioceptive and sensory signals and a self-organizing... more
We have developed an 8 d.o.f. robot hand which has been tested with three computational models of haptic perception. Two of the models are based on the tensor product of different proprioceptive and sensory signals and a self-organizing map (SOM), and one uses a novel self-organizing neural network, the T-MPSOM. The computational models have been trained andtested with a set
Research Interests:
We investigate four different methods for background estimation in calcium imaging of the insect brain and evaluate their performance on six data sets consisting of data recorded from two sites in two species of moths. The calcium... more
We investigate four different methods for background estimation in calcium imaging of the insect brain and evaluate their performance on six data sets consisting of data recorded from two sites in two species of moths. The calcium fluorescence decay curve outside the potential response is estimated using either a low-pass filter or constant, linear or polynomial regression, and is subsequently used to calculate the magnitude, latency and duration of the response. The magnitude and variance of the responses that are obtained by the different methods are compared, and, by computing the receiver operating characteristics of a classifier based on response magnitude, we evaluate the ability of each method to detect the stimulus type and conclude that a polynomial approximation of the background gives the overall best result.
Research Interests:
This paper outlines the epigenetic logic of the emergence and elaboration of symbolization. The account is based upon considerations arising from the study of the ontogenesis and phylogenesis of symbolic communication, abstraction from... more
This paper outlines the epigenetic logic of the emergence and elaboration of symbolization. The account is based upon considerations arising from the study of the ontogenesis and phylogenesis of symbolic communication, abstraction from which yields generalizations regarding the necessary developmental pre-requisites for the capacity for symbolization and the processes of the elaboration of symbolic capacity into language.
IntroductionA visually orienting mobile robot must cope with anumber of changes to its environment. Mostimportantly, it must be able to identify its locationeven when objects in the environment have beenmoved or when the illumination... more
IntroductionA visually orienting mobile robot must cope with anumber of changes to its environment. Mostimportantly, it must be able to identify its locationeven when objects in the environment have beenmoved or when the illumination conditions havechanged. We report experiments with a template-basedself-localization method that operates in real time on aPentium 133 MHz PC (Balkenius and Kopp 1996a, b;1997). The algorithm
Research Interests:
A mobile robot navigating in an unstructured environment faces many difficult problems for which vision may potentially offer useful solutions. The XT-1 (eXpectation based Template matching) architecture was developed in an attempt to... more
A mobile robot navigating in an unstructured environment faces many difficult problems for which vision may potentially offer useful solutions. The XT-1 (eXpectation based Template matching) architecture was developed in an attempt to address many of these problems with similar constructions. The current system handles such diverse problems as landmark and place recognition, the generation of orienting and anticipatory saccades,
Research Interests:
IntroductionThe abilities of the currently existing hand prostheses are typically limited toopening or closing the hand. This limits the usefulness of the prosthesisconsiderably compared to the many degrees of freedom in an intact hand.... more
IntroductionThe abilities of the currently existing hand prostheses are typically limited toopening or closing the hand. This limits the usefulness of the prosthesisconsiderably compared to the many degrees of freedom in an intact hand. In order todevelop more advanced hand prostheses two main problems have to be solved. Thefirst is to develop more advanced mechanical solutions that allows for more
Research Interests:
This paper shows how a representation at multiple scales can be used for generalization in instrumental learning. The use of such representations is an efficient way to code similarities and differences within a stimulus dimension, and... more
This paper shows how a representation at multiple scales can be used for generalization in instrumental learning. The use of such representations is an efficient way to code similarities and differences within a stimulus dimension, and allows a learning system to generalize easily between various situations. A neural network based model of classical and instrumental conditioning is presented and its
Research Interests:
Research Interests:
Research Interests:
Persons with autism, probably due to early sensory impairments, attend to and select for stimuli in an uncommon way. Inhibition of some features of a stimulus, such as location and shape, might be intact, while other features are not as... more
Persons with autism, probably due to early sensory impairments, attend to and select for stimuli in an uncommon way. Inhibition of some features of a stimulus, such as location and shape, might be intact, while other features are not as readily inhibited, for example color. Stimuli irrelevant to the task might be attended to. This results in a learning process
Research Interests:
Research Interests:
ABSTRACT We investigating the role of anticipation and attention in a dynamic environment in a number of large scale simulations of an agent that tries to negotiate a number of gates that continuously open and close. In particular we have... more
ABSTRACT We investigating the role of anticipation and attention in a dynamic environment in a number of large scale simulations of an agent that tries to negotiate a number of gates that continuously open and close. In particular we have looked at learning mechanisms that can predict the future positions of the gates and control strategies that will allow the agent to pass through the gates unharmed. The simulations reported below use the AARC architecture [1]. This architecture combines a large number of different cognitive mechanisms. In Experiment 1, the task for the agent is to pass through a single gate and in Experiment 2, to pass through three successive gates. The results shows that the AARC architecture is flexible enough to handle very diverse situations. It is also somewhat surprising that linear predictors are sufficient in most cases.
ABSTRACT A novel model of role of conditioning in attention is presented and evaluated on a Nao humanoid robot. The model implements conditioning and habituation in interaction with a dynamic neural field where different stimuli compete... more
ABSTRACT A novel model of role of conditioning in attention is presented and evaluated on a Nao humanoid robot. The model implements conditioning and habituation in interaction with a dynamic neural field where different stimuli compete for activation. The model can be seen as a demonstration of how stimulus-selection and action-selection can be combined and illustrates how positive or negative reinforcement have different effects on attention and action. Attention is directed toward both rewarding and punishing stimuli, but appetitive actions are only directed toward positive stimuli. We present experiments where the model is used to control a Nao robot in a task where it can select between two objects. The model demonstrates some emergent effects also observed in similar experiments with humans and animals, including attentional blocking and latent inhibition.
Research Interests:
... e Telematica, Universita' di Genova, Via all'Opera Pia 13 - 16145 Genova, Italy 3 Lund University Cognitive Science, Kungshuset, Lundagård SE - 222 22 Lund, Sweden dimitri.ognibene@istc. cnr ... Whitehead, SD,... more
... e Telematica, Universita' di Genova, Via all'Opera Pia 13 - 16145 Genova, Italy 3 Lund University Cognitive Science, Kungshuset, Lundagård SE - 222 22 Lund, Sweden dimitri.ognibene@istc. cnr ... Whitehead, SD, Ballard, DH: Learning to perceive and act by trial and error. Mach. ...
Research Interests:
How can visual selective attention guide eye movements so as to collect information and identify targets potentially relevant for action? Many models have been proposed that use the statistical properties of images to create a dynamic... more
How can visual selective attention guide eye movements so as to collect information and identify targets potentially relevant for action? Many models have been proposed that use the statistical properties of images to create a dynamic bottom-up saliency map used to guide saccades to potentially relevant locations. Since the concept of saliency map was introduced, it has been incorporated in
Research Interests: Selective Attention, Reinforcement Learning, Visual attention, Neural Network, Information Flow, and 12 moreObject Recognition, Robot Arm, Instrumental Conditioning, Saliency Map, Top Down Processing, Learning Process, Statistical Properties, Eye Movement, Short Term Memory, Mechanism of action, Bottom Up, and Top Down
Research Interests:
Research Interests:
The paper presents a general architecture for behaviour based control systems for autonomous agents. A number of archi tectural principles are proposed which make it possible to combine reactive control with learning and problem solving... more
The paper presents a general architecture for behaviour based control systems for autonomous agents. A number of archi tectural principles are proposed which make it possible to combine reactive control with learning and problem solving in a coherent way. In particular, I investigate the interaction between reinforcement learning, internal world models and dynamic action selection as well as a number
Research Interests:
Research Interests:
This book has provided various theoretical perspectives on anticipatory processes in natural and artificial cognitive systems. Advantages have been proposed and confirmed in various detailed case studies, which may have given the reader... more
This book has provided various theoretical perspectives on anticipatory processes in natural and artificial cognitive systems. Advantages have been proposed and confirmed in various detailed case studies, which may have given the reader detailed insights into anticipatory processes and their importance in various cognitive systems tasks. To wrap up these advantages and give a concluding overview of various current anticipatory process advantages, this final chapter highlights a concise collection of precise success stories of ...
Research Interests:
Research Interests:
ABSTRACT Although attention can be purely reactive, like when we react to an unexpected event, in most cases, attention is under deliberate control anticipating events in the world. Directing attention and preparing for action takes time,... more
ABSTRACT Although attention can be purely reactive, like when we react to an unexpected event, in most cases, attention is under deliberate control anticipating events in the world. Directing attention and preparing for action takes time, and it is thus useful to be able to predict where an important event will occur in the environment and direct attention to it even before it happens. Another reason for the need for anticipation is the processing delays in the visuomotor system. In the human system it takes at least 100 ms to detect a visual target (Lamme and Roelfsema, 2000) and to just look at a moving object, we need to anticipate its movement to control the muscles of the eyes to move our gaze to the location where the target will be (von Hofsten and Rosander, 1997). The role of anticipation in attention can also be seen in the close connection between attention and action (Balkenius, 2000).
The moth Macroglossum stellatarum can learn the colour and sometimes the odour of a rewarding food source. We present data from 20 different experiments with different combinations of blue and yellow artificial flowers and the two odours... more
The moth Macroglossum stellatarum can learn the colour and sometimes the odour of a rewarding food source. We present data from 20 different experiments with different combinations of blue and yellow artificial flowers and the two odours honeysuckle and lavender. The experiments show that learning about the odours depends on the colour used. By training on different colour-odour combinations and testing on others, it becomes possible to investigate the exact relation between the two modalities during learning. Three computational models were tested in the same experimental situations as the real moths and their predictions were compared to the experimental data. The average error over all experiments as well as the largest deviation from the experimental data were calculated. Neither the Rescorla-Wagner model or a learning model with independent learning for each stimulus component were able to explain the experimental data. We present the new categorisation model, which assumes that the moth learns a template for the sensory attributes of the rewarding stimulus. This model produces behaviour that closely matches that of the real moth in all 20 experiments.
Research Interests:
Research Interests:
We describe an efficient algorithm for visual landmark recognition based on elastic template matching. Each landmark is represented by a set of features and their spatial relations. The matching procedure allows for visual transformations... more
We describe an efficient algorithm for visual landmark recognition based on elastic template matching. Each landmark is represented by a set of features and their spatial relations. The matching procedure allows for visual transformations such as translation and expansion, or even partial occlusion. Robot navigation in a natural environment calls for a sophisticated sensory system. In this paper we propose a method for visual landmark recognition based on elastic template matching that can be used for both positioning and navigation. We define positioning as a process that uses the sensory input of the robot to derive the location and orientation of the robot with respect to a map of the environment. Navigation, on the other hand, is the use of external cues and internal structures to derive a path through the environment. We will mainly be concerned with navigation, but the methods developed below can also be used for pure positioning. The landmark recognition method we describe be...