University of Sheffield
An investigation into the utility of modern computer graphics methods for insect vision research
Abstract
dc:description.abstractDespite their small size, insects retain the ability to expertly navigate complex terrain, integrating visual cues for homing, foraging and discrimination tasks. In a world dominated by computationally-intensive visual processing algorithms, not only biologists and ethologists, but also roboticists seek to understand the seemingly unparalleled efficiency gains found in the neurological and physiological configurations of these animals. Key to insect’s vision is their use of compound eyes that are – unlike mammalian eyes – composed of hundreds to tens of thousands of individual light-sensing and focusing optical arrangements called ommatidia. Ommatidia provide the insect with a low-resolution view of the world, exhibiting a myriad of evolutionary adaptions based on ecological context. A predominant methodology within neuroethology and biomimetic robotics for the study of compound vision is the use of compound eye models (CEMs) to simulate the visual perspective of these insects. Historically, these CEMs have formed the back-bone of experiments studying the features of compound vision, such as its wide field of view or low and non-uniform spatial resolution. However, these CEMs are often limited, with assumptions or simplifications of the eye design under test made in order to expedite the experimental process. Software that has been designed to accurately model the light received at compound eyes is often slow and unable to operate at scale. To quantitatively observe trends across large spatial scales similar to those travelled by real-world insects, a method to recreate and evaluate the visual information available to these animals both accurately and at speed is required. This work presents a data-backed methodology to perform these large-scale experiments, using state-of-the-art ray-tracing technologies to integrate realistic compound-eye morphological data, 3D captures of real-world locations and insect behaviour to perform analysis of the visual information available to insects while navigating at scale, demonstrating the utility of this high-throughput approach.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Sheffield
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Millward, Blayze
- Advisors dc:contributor.advisor
-
- Mangan, Michael
- Maddock, Steve
- ORCID person:identifier.orcid
Rights
- Language dc:language.dcterms_iso639_2
- eng