Bionic navigation: From emergent retinal optic flow to locomotor ballistic corrections in humans, for robotics
Uloženo v:
| Název: | Bionic navigation: From emergent retinal optic flow to locomotor ballistic corrections in humans, for robotics |
|---|---|
| Autoři: | Nguyen, Björnborg, 1992 |
| Zdroj: | Closed raw data set for: Intermittent control and retinal optic flow when maintaining a curvilinear path. |
| Témata: | Bionics, active fixation, neuromuscular behavior, smooth pursuit, robotic navigation, visual perception, human locomotor behavior, active gazing, optic flow, locomotor control, retinal optic flow |
| Popis: | This thesis explores the development of bio-inspired methods for addressing complex navigation tasks in cyber-physical robotic systems. In human locomotion, a sensation of a visual flow cue is created by continuously registering moving visual features on the human retina. An interpretation of visual flow cues forms a low-level motion perception more commonly known as retinal optic flow . It is often mentioned and credited in recent human locomotor research, but is limited as a theory or concept. A computational method of reconstructing retinal optic flow fields in humans is introduced and studied along with its effect on intermittent control to initiate informed ballistic corrections . Unifying the theories of human visual perception, intermittent control, muscular control, and coordination, the analysis from experiments reveals that human response time is approximately 0.14s. To achieve retinal optic flow reconstruction for robotics, a set of optic flow estimators is fairly and systematically evaluated on the run-time performance, reliability, and accuracy criteria. This work developed a formalized methodology utilizing a microservice paradigm and containerization technology to perform benchmarking and generate results. Doing so enables continuous integration, continuous deployment, and continuous experimentation, which are beneficial for research and development. Furthermore, this thesis found that the readiness of vehicles for adopting modern robotic software, with special emphasis on real-time computing, has matured and essentially turned vehicles into mobile data centers and capable robots. To show this partially, two optic flow-based local localization methods are demonstrated for marine vehicles in a littoral environment. |
| Popis souboru: | electronic |
| Přístupová URL adresa: | https://research.chalmers.se/publication/547987 https://research.chalmers.se/publication/547987/file/547987_Fulltext.pdf |
| Databáze: | SwePub |
Nájsť tento článok vo Web of Science | Abstrakt: | This thesis explores the development of bio-inspired methods for addressing complex navigation tasks in cyber-physical robotic systems. In human locomotion, a sensation of a visual flow cue is created by continuously registering moving visual features on the human retina. An interpretation of visual flow cues forms a low-level motion perception more commonly known as retinal optic flow . It is often mentioned and credited in recent human locomotor research, but is limited as a theory or concept. A computational method of reconstructing retinal optic flow fields in humans is introduced and studied along with its effect on intermittent control to initiate informed ballistic corrections . Unifying the theories of human visual perception, intermittent control, muscular control, and coordination, the analysis from experiments reveals that human response time is approximately 0.14s. To achieve retinal optic flow reconstruction for robotics, a set of optic flow estimators is fairly and systematically evaluated on the run-time performance, reliability, and accuracy criteria. This work developed a formalized methodology utilizing a microservice paradigm and containerization technology to perform benchmarking and generate results. Doing so enables continuous integration, continuous deployment, and continuous experimentation, which are beneficial for research and development. Furthermore, this thesis found that the readiness of vehicles for adopting modern robotic software, with special emphasis on real-time computing, has matured and essentially turned vehicles into mobile data centers and capable robots. To show this partially, two optic flow-based local localization methods are demonstrated for marine vehicles in a littoral environment. |
|---|