Attention mechanisms are the fundamental processes in biological systems, responsible for prioritizing the elements of the visual scene to be attended, i.e., to control perceptual resources and cope with the brain computational limitations. Humans, for instance, rely on space-variant sensing (foveal vision), and on stimulus-driven (bottom-up) and goal-driven (top-down) information processing mechanisms to define where in the visual input the attentional foci should be oriented to. This way, information processing is constrained and directed towards salient or task-relevant stimuli.
Likewise, an important issue in many computer vision applications requiring real-time performance, resides in the involved computational effort, especially in robotics where energy efficient, fast and accurate perception is a fundamental requirement, e.g., in visual localization and servoing during grasping, manipulation and hand-over of tools to human or machine collaborators. In humanoid robotics, in particular, real-time operation is conditioned by physical limitations on on-board computational and power resources, as well as sensory data transmission bandwidth.
State-of-the-art artificial systems are being deployed successfully in complex unstructured and human populated environments. However, most systems rely on expensive and powerful sensors and computational hardware apparatus, to ensure safe and real-time autonomous operation. Unfortunately, these constraints typically compromise their degree of autonomy, run-time duration and size.
Therefore, the aim of this Research Topic is to seek contributions that delve beyond the current state-of-the-art in resource-constrained computer vision solutions, ranging from the sensory level to higher-level cognitive functionalities, for various visual understanding tasks, including but not limited to recognition, volumetric and semantic reconstruction, pose estimation, localization, tracking and mapping.
Authors are encouraged to submit contributions with demonstrated applicability in unmanned robotics systems, i.e., from ground, aerial, underwater, bio-inspired. In particular real-world use-cases – e.g., from micro-aerial inspection vehicles to social assistive humanoid robots – where efficient and low-latency perception is required to cope with computational, power and real-time application requirements.
This Research Topic aims to collect submissions which demonstrate how vision can be enhanced with novel and especially bio-inspired technologies. We are seeking contributions on the following topics of interest, but not limited to:
• Efficient perception with unconventional vision sensors, such as event cameras, focal-plane sensor-processor, software/hardware retinas, foveal vision, plenoptic cameras
• Space and time-variant visual attention and constrained resource allocation computational mechanisms for artificial systems with limited resources
• Efficient neural network architectures and learning visual mechanisms on resource-constrained robots
• Biologically principled models for active vision
• Biologically motivated approaches to vision for embedded systems with hard real-time constraints
Topic Editor Rui Pimentel de Figueiredo is employed by Capra Robotics ApS. All other Topic Editors declare no competing interests with regards to the Research Topic subject.
Attention mechanisms are the fundamental processes in biological systems, responsible for prioritizing the elements of the visual scene to be attended, i.e., to control perceptual resources and cope with the brain computational limitations. Humans, for instance, rely on space-variant sensing (foveal vision), and on stimulus-driven (bottom-up) and goal-driven (top-down) information processing mechanisms to define where in the visual input the attentional foci should be oriented to. This way, information processing is constrained and directed towards salient or task-relevant stimuli.
Likewise, an important issue in many computer vision applications requiring real-time performance, resides in the involved computational effort, especially in robotics where energy efficient, fast and accurate perception is a fundamental requirement, e.g., in visual localization and servoing during grasping, manipulation and hand-over of tools to human or machine collaborators. In humanoid robotics, in particular, real-time operation is conditioned by physical limitations on on-board computational and power resources, as well as sensory data transmission bandwidth.
State-of-the-art artificial systems are being deployed successfully in complex unstructured and human populated environments. However, most systems rely on expensive and powerful sensors and computational hardware apparatus, to ensure safe and real-time autonomous operation. Unfortunately, these constraints typically compromise their degree of autonomy, run-time duration and size.
Therefore, the aim of this Research Topic is to seek contributions that delve beyond the current state-of-the-art in resource-constrained computer vision solutions, ranging from the sensory level to higher-level cognitive functionalities, for various visual understanding tasks, including but not limited to recognition, volumetric and semantic reconstruction, pose estimation, localization, tracking and mapping.
Authors are encouraged to submit contributions with demonstrated applicability in unmanned robotics systems, i.e., from ground, aerial, underwater, bio-inspired. In particular real-world use-cases – e.g., from micro-aerial inspection vehicles to social assistive humanoid robots – where efficient and low-latency perception is required to cope with computational, power and real-time application requirements.
This Research Topic aims to collect submissions which demonstrate how vision can be enhanced with novel and especially bio-inspired technologies. We are seeking contributions on the following topics of interest, but not limited to:
• Efficient perception with unconventional vision sensors, such as event cameras, focal-plane sensor-processor, software/hardware retinas, foveal vision, plenoptic cameras
• Space and time-variant visual attention and constrained resource allocation computational mechanisms for artificial systems with limited resources
• Efficient neural network architectures and learning visual mechanisms on resource-constrained robots
• Biologically principled models for active vision
• Biologically motivated approaches to vision for embedded systems with hard real-time constraints
Topic Editor Rui Pimentel de Figueiredo is employed by Capra Robotics ApS. All other Topic Editors declare no competing interests with regards to the Research Topic subject.
