Dynamic Neural Fields as Building Blocks of a Cortex-Inspired Architecture for Robotic Scene Representation

Zibner, Stephan K. U.; Faubel, Christian; Iossifidis, Ioannis; Schöner, Gregor

doi:10.1109/tamd.2011.2109714

Cited by 37 publications

(35 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Even though this framework might be controversial to the understanding of biological systems, it has been successfully applied in robotic design (Metta et al, 2008;Zibner et al, 2011). This approach conceives cognitive agents as managers of a versatile neural architecture consisting of coupled dynamical systems in flexible interaction.…”

Section: Neural and Behavioral Dynamicsmentioning

confidence: 99%

Sensory flow shaped by active sensing: sensorimotor strategies in electric fish

Hofmann

Sanguinetti-Scheck

Künzel

et al. 2013

Journal of Experimental Biology

View full text Add to dashboard Cite

SummaryGoal-directed behavior in most cases is composed of a sequential order of elementary motor patterns shaped by sensorimotor contingencies. The sensory information acquired thus is structured in both space and time. Here we review the role of motion during the generation of sensory flow focusing on how animals actively shape information by behavioral strategies. We use the well-studied examples of vision in insects and echolocation in bats to describe commonalities of sensory-related behavioral strategies across sensory systems, and evaluate what is currently known about comparable active sensing strategies in electroreception of electric fish. In this sensory system the sensors are dispersed across the animalʼs body and the carrier source emitting energy used for sensing, the electric organ, is moved while the animal moves. Thus ego-motions strongly influence sensory dynamics. We present, for the first time, data of electric flow during natural probing behavior in Gnathonemus petersii (Mormyridae), which provide evidence for this influence. These data reveal a complex interdependency between the physical input to the receptors and the animalʼs movements, posture and objects in its environment. Although research on spatiotemporal dynamics in electrolocation is still in its infancy, the emerging field of dynamical sensory systems analysis in electric fish is a promising approach to the study of the link between movement and acquisition of sensory information.

show abstract

Section: Neural and Behavioral Dynamicsmentioning

confidence: 99%

Sensory flow shaped by active sensing: sensorimotor strategies in electric fish

Hofmann

Sanguinetti-Scheck

Künzel

et al. 2013

Journal of Experimental Biology

View full text Add to dashboard Cite

show abstract

“…Modern robotics has also embraced biological algorithms, with biologically-inspired SLAM algorithms [23,24], obstacle avoidance, and complete robotic architectures [25]. Attempts to build a comprehensive biologically plausible system have focussed on dynamic field-based models of different aspects of vision [26] and cognitive robots [25]. A good comparative summary of computer vision and biological vision is given in [27].…”

Section: Related Workmentioning

confidence: 99%

“…Each object is updated as the scene changes. In the near future, we will extend this simple model with a biologically motivated dynamical-field represenation [26]. For autonomous robots, we also use a dynamical 2D spatial map of obstacles which is actively updated (using reinforcement learning) and fades with time (see Fig.…”

Section: Long-term Memory and High-level Reasoningmentioning

confidence: 99%

Biological Models for Active Vision: Towards a Unified Architecture

Terzic

Lobato

Saleiro

et al. 2013

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Building a general-purpose, real-time active vision system completely based on biological models is a great challenge. We apply a number of biologically plausible algorithms which address different aspects of vision, such as edge and keypoint detection, feature extraction, optical flow and disparity, shape detection, object recognition and scene modelling into a complete system. We present some of the experiments from our ongoing work, where our system leverages a combination of algorithms to solve complex tasks.

show abstract

“…However, vision is a complex interplay between bottom-up and top-down processes, and attention is also driven by high-level expectations [10,15], which is why it should be possible for high-level concepts about colour and shape to determine what is salient in an image. This is the approach often taken in cognitive robotics, where there is tight coupling between feature representation and action [27]. We believe that powerful scene interpretation can be built on top of interpretable, semantically meaningful features, which can simplify top-down queries because they directly relate to higher-level descriptions.…”

Section: Introductionmentioning

confidence: 99%

Interpretable Feature Maps for Robot Attention

Terzic

Buf

2017

Universal Access in Human–Computer Interaction. Design and Development Approaches and Methods

View full text Add to dashboard Cite

Abstract. Attention is crucial for autonomous agents interacting with complex environments. In a real scenario, our expectations drive attention, as we look for crucial objects to complete our understanding of the scene. But most visual attention models to date are designed to drive attention in a bottom-up fashion, without context, and the features they use are not always suitable for driving top-down attention. In this paper, we present an attentional mechanism based on semantically meaningful, interpretable features. We show how to generate a low-level semantic representation of the scene in real time, which can be used to search for objects based on specific features such as colour, shape, orientation, speed, and texture.

show abstract

Dynamic Neural Fields as Building Blocks of a Cortex-Inspired Architecture for Robotic Scene Representation

Cited by 37 publications

References 57 publications

Sensory flow shaped by active sensing: sensorimotor strategies in electric fish

Sensory flow shaped by active sensing: sensorimotor strategies in electric fish

Biological Models for Active Vision: Towards a Unified Architecture

Interpretable Feature Maps for Robot Attention

Contact Info

Product

Resources

About