András Róka

An emotional engine model inspired by human-dog interaction

We use an engineering approach based on state space modeling to develop the emotional engine of an artificial agent. The model is inspired by the latest achievements of human-dog interaction research. Our goal is to build a model that can be easily integrated into existing applications in the field of robotics, computing and communication systems. The proposed model ensures this integration by using simple arithmetics, while being scalable and stable.

Contrast vision-based grasp force feedback in telemanipulation

The operator usually uses his hands to perform manipulation tasks. She has to feel the reaction forces in high-precision manipulation. The mechanism applied in the human interface device for force and tactile feedback is the most challenging (and usually the most expensive) part of a telepresence device. Bilateral control is a common solution in these systems. However, the coupled force and position control loops have contradicting requirements. Usually, the human operator has to learn a lot until she can ”feel the task” in her hand, even in direct manipulation. However, using the plasticity of the human brain sensory substitution can be used to achieve a cheaper and partially decoupled system. The main contribution of this paper is a model of human contrast vision, which is applied to feedback of the grasping force in a telemanipulation system in order to decouple the bilateral control and to achieve a less complex haptic device. A contrast model of a the human vision is introduced and it is applied for grasping force feedback via peripheral vision. The preliminary results are presented.

Building a human-dog interaction inspired emotional engine model

We propose a state space based engineering solution for the emotional engine model of an artificial agent. Our model takes its inspirations from the evolution theory and the latest achievements of human-dog interaction research. We present our emotional model together with its background, show our simulation results and present the ongoing work concerning the applications that we chose to be extended by our engineering solution.

Visual Cortex Inspired Vertex and Corner Detection

Sets of corners and crossings in images characterize the shape and position of visual objects. The detection of these features not only supports, but is also a necessary condition to successful object recognition in human visual perception. Results in cognitive research have shown that the visual cortex has a mechanism for the detection of corners and crosses, and makes use of this in higher-level visual processing. In this paper, we propose an artificial neural architecture as a possible model for corner- and cross-sensitive cells - as an extension to the VFA (visual feature array) module. The model assumes a hierarchical structure of node-, endpoint-, crossand corner-sensitive cells. Through test results, the benefits of such a hierarchical structure is shown in terms of higher-level object recognition

Contrast sensitivity model of the human eye

In this paper, we propose a mathematical model of human photopic contrast sensitivity. The model is based on a novel functional block diagram, focusing more on the information theoretical nature of human contrast sensitivity. At the same time, several other aspects of human vision are also considered. In order to obtain an estimate of the nonlinear projection of the retinal image, the numerical results of Drasdo and Fowler were used. Cone density measurements along the major meridians of the retina, performed by Curcio & Allen, were also used. Retinal midget ganglion cell receptive field density was extended to the far periphery, based on the results of Drasdo and colleagues and Curcio & Allen. Considering cell densities and optical properties of the human eye, the low-pass filtering components of human contrast sensitivity have been characterized by a simple but adequate mathematical formula. The high-pass filtering of neural origin was also estimated based on midget receptive field densities as well as other experimental results. Further models are presented for the calculation of photon noise, neural noise and spatial integration, according to the previous results in the literature. The model presented in this paper has been validated by several experimental measurements, both for foveal and peripheral vision, at several luminance levels and experimental set-ups. Results were shown to be better fitting than the model of Barten [2] and Rovamo et al [37].

Visual feedback techniques for telemanipulation and system status sensualization

A general effort in telemanipulation is to establish as many and as high quality sensory channels as possible between the human operator and the remote site. However, most of the tasks require only a portion of human sensory skills, typically - besides visual information - kinesthetic, haptic and tactile force-feedback. But realizing teleoperators providing transparent force-feedback over a long distance is still a challenge because of the low bandwidth, varying network delay and the probably unknown remote environment. Therefore, the senses that are not necessary for controlling the process should be used to support the operator to enhance or substitute the force feedback. In system status sensualization the goal is to provide an interpretation of sensory information that can be easily and effectively perceived by the operator. However, traditional analogous or digital instrumentation requires that the operator directs his sight right to the instrument to read the values. Furthermore, these values are usually important only if they are in an abnormal state. In case of a quite complex system (an airplane or a power plant) this process is very time-consuming and often needs several operators. In this paper we propose new visualization methods using peripheral vision to transmit visual information matched to the capabilities of the human visual system. The novel method could be applied for both sensory substitution and system status sensualization. The theoretical background of this work is a mathematical model of the human photopic contrast sensitivity proposed by the authors in a previous work. The novel visualization methods are analyzed experimentally via comparative tests.

Cognitive Informatics based DIND for Corner and Crossing Detection in Intelligent Space

Intelligent space is a space of distributed sensory intelligence and actuators. The basic component of intelligent space is the distributed intelligent network device (DIND), responsible for intelligent sensing and estimation. It is important for intelligent space to be aware of state of its internal environment, such as the class and position of objects it contains. The ability of object recognition is thus necessary in intelligent space. Existing cognitive computational systems like the mammalian cortex perform extremely well when it comes to visual object recognition. Psychological experiments have shown that image features such as edges and especially corners are of great importance in cognitive object recognition. Inspired by cognitive recognition systems, the proposed cognitive informatics model addresses the problem of vertex and corner detection. The model integrates the previously developed visual feature array (VFA), which is a cognitive model of oriented edge detection. The ultimate goal of the presented model is to provide suitable input for a cognitive object recognition system

Computational model of the human eye for intelligent systems

The knowledge of the spatiotemporal sensitivity of the human eye can be essential for those practical applications where the communication of an informatical system and a human subject can be crucial (topics: cognitive informatics, cognitive infocommunication), such as intelligent video-coding, driving assistant systems or intelligent instrumental boards. In this paper we will present the spatiotemporal sensitivity of the human eye for achromatic stimuli at photopic light levels. To our knowledge, it is the first comprehensive model that: (1) considers the spatiotemporal transfer function of P and M ganglion cells along the retina, (2) contains the fast adaptation mechanisms (contrast gain control and luminance gain control) and (3) discusses the role of the receptor and post-receptor light adaptation separately.

Modeling awareness of surface geometry in auditory cognitive channels

Abstract We refer to channels of communication which link the user to various electronic appliances and computers as cognitive communication channels. One especially interesting research topic related to cognitive communication channels deals with a special application called sensory substitution, when information is conveyed through a channel other than the one that is normally used. Besides offering a glimpse of hope to those living with injured sensory organs, sensory substitution can have benefits when designing user interfaces in terms of cost-effectiveness, the reduction of network delays, as well as the prevention of channel overloading. One such application in robot control could be the substitution of haptic feedback gloves (which are costly, bulky and have relatively few vibration states) using parameter-rich audio interfaces. In this paper, we experiment with a model capable of representing geometrical properties of surfaces in terms of audio-substituted tactile percepts, while at the same time modeling the more subjective dimension of user awareness.