Abstract
Fast reaction to sudden and potentially interesting stimuli is a crucial feature for safe and reliable interaction with the environment. Here we present a biologically inspired attention system developed for the humanoid robot iCub. It is based on input from unconventional event-driven vision sensors and an efficient computational method. The resulting system shows low-latency and fast determination of the location of the focus of attention. The performance is benchmarked against an instance of the state of the art in robotics artificial attention system used in robotics. Results show that the proposed system is two orders of magnitude faster that the benchmark in selecting a new stimulus to attend.
Highlights
For successfully interacting with the environment in daily tasks, it is crucial to quickly react to ubiquitous dynamic stimuli
In this manuscript we described Event-Driven Visual Attention (EVA), a real-time implementation of selective attention based on the bio-inspired model proposed in the foundational work of Itti and Koch (2001), that uses a frame-less asynchronous vision sensor as input
The overall performance of the developed system takes advantage of the efficient information encoding operated of the sensor, its high dynamic range, low response latency and high temporal resolution
Summary
For successfully interacting with the environment in daily tasks, it is crucial to quickly react to ubiquitous dynamic stimuli. An alternative approach is the implementation of simplified models of attention systems based on frame-less event-driven neuromorphic vision sensors, so far realized with the design of ad hoc dedicated hardware devices (Bartolozzi and Indiveri, 2009; Sonnleithner and Indiveri, 2012) Along this line of research, we developed an event-driven, attention system capable of selecting interesting regions of the visual input with a very short latency. Intermediate hybrid approaches can be implemented by reconstructing frames from the events and applying the vast collection of available standard machine vision algorithms This approach would suffer from errors in the frame reconstruction due to drifts in the gray level calculation, it would increase the latency of the response and loose the temporal resolution gained by the use of event-driven sensors, hindering the full exploitation of the neuromorphic approach advantages. We will describe EVA, show the improved latency in the selection of salient stimulus and compare its performance with the well-known state-of-the art frame-based selective attention system from the iLab Neuromorphic Vision C++ Toolkit (iNVT), 1 developed at the University of Southern California
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
