Neuroscience Research in Spatial Navigation Using Robotic Animals
| dc.contributor.author | Sutton, Nate | |
| dc.date.accessioned | 2020-05-12T20:10:21Z | |
| dc.date.available | 2020-05-12T20:10:21Z | |
| dc.date.issued | 2020 | |
| dc.description.abstract | The focus of this review is neuroscience research using robots programmed to navigate spatial environments using methods based on animal cognition. The use of cognition-based methods in robot navigation has multiple translational applications such as self-driving cars, and robots that handle hostile, dull, dangerous, and dirty activities. The use of robots for testing spatial cognition neural theories can provide advantages in accumulating real-world environment data, obtaining a diversity of test environments, and being able to directly reuse test environments where animal neural recordings were captured. Multiple types of neurons encode separate properties of space that are used in navigation activities, e.g., using information around one’s body to determine position while comparing that to a goal path (path integration). Research producing better knowledge of brain spatial processing mechanisms with those cell types helps inform the design of algorithms for navigation in robots. A type of advance that has been made in neural algorithm design was the inclusion of conjunctive cells modeling. The cells combine and compress sensory information in cells, e.g., visual landmarks, odors, sounds, touch, for neural computational efficiency gains. A type of challenge in navigation is simultaneous localization and mapping (SLAM), which is a problem that has attracted widespread attention in the field of computer vision. An innovative approach toward SLAM that is based on rodent cognition is named RatSLAM. It uses an attractor network algorithm to process spatial information cues to understand environments, allowing the navigation to be robust to ambiguous spatial landmark information and path integration errors. RatSLAM has been expanded on with features such as merging information between robot rats toward their common objective of understanding spatial environments. Integration of advanced forms of neuromorphic hardware, for instance, International Business Machines’ (IBM) Neurosynaptic System, into robots designed for navigation enables computational efficiency gains. Research with robots designed with navigation algorithms based on neural cognition has been advanced by studies investigating multiple factors involved in the activity of navigation. Several research projects have used discoveries learned from the fields of robotic navigation and neuroscience to create new investigations that improved the state of spatial navigation science. | |
| dc.identifier.uri | https://hdl.handle.net/1920/11761 | |
| dc.language.iso | en_US | |
| dc.rights | Attribution-ShareAlike 3.0 United States | |
| dc.rights.uri | https://creativecommons.org/licenses/by-sa/3.0/us/ | |
| dc.subject | Spatial Memory | |
| dc.subject | Neural Networks | |
| dc.subject | Grid Cells | |
| dc.title | Neuroscience Research in Spatial Navigation Using Robotic Animals | |
| dc.type | Other |