SEMF | Anthony Cohn | Spatial Intelligence and Challenges of the Spatial World @SEMF | Uploaded November 2022 | Updated October 2024, 17 hours ago.
Talk kindly contributed by Anthony G. Cohn in SEMF's 2022 Spacious Spatiality
semf.org.es/spatiality
TALK ABSTRACT
Agents, whether human or robotic, inhabit a spatial world, whether the real world or a virtual one; they observe it, the events that take place in it, and move around in it. They need to navigate in it, perform tasks which will affect the positions and other aspects of objects in the world. They have goals to achieve and may need to communicate about the world they inhabit with other agents, without necessarily having complete knowledge about the world. In this session, I will talk about why reasoning about space is important; what makes space “special”; the variety of spatial tasks an agent may face; and what are the particular challenges of endowing an agent with spatial intelligence. I will talk about some of the techniques the have been developed to endow agents with spatial intelligence. Amongst these are methods involving qualitative spatial reasoning which has been a particular focus of mine over the years. The real challenge is how to enable an agent to make commonsense spatial inferences, and what is the role of natural language in relation to commonsense spatial reasoning?
TALK MATERIALS
· A Survey of Qualitative Spatial Representations: eprints.whiterose.ac.uk/83881
· Automated Commonsense Spatial Reasoning: Still a Huge Challenge: eprints.whiterose.ac.uk/172683
ANTHONY G. COHN
School of Computing, Leeds University: eps.leeds.ac.uk/computing
Leeds University profile: eps.leeds.ac.uk/computing/staff/76/professor-anthony-tony-g-cohn-freng-ceng-citp
ResearchGate:researchgate.net/profile/Anthony-Cohn-2
Google Scholar: scholar.google.com/citations?user=tal4mMkAAAAJ
LinkedIn: linkedin.com/in/tonycohn
SEMF NETWORKS
Website: semf.org.es
Twitter: twitter.com/semf_nexus
LinkedIn: linkedin.com/company/semf-nexus
Instagram: instagram.com/semf.nexus
Facebook: facebook.com/semf.nexus
MUSIC
youtube.com/user/Baroquenoise
00:00:12 Introduction
00:00:56 Anthony’s background
00:03:01 SPATIAL DATA AND KNOWLEDGE
00:03:02 Ubiquity of spatial data across disciplines
00:04:10 Spatial intelligence in animals
00:05:38 Overview of the talk
00:06:10 What’s special about spatial?
00:08:47 The spatial problem solving situation
00:11:51 Qualitative spatial knowledge
00:15:10 Spatial language
00:18:48 Acquiring spatial data and knowledge
00:22:29 Spatial tasks
00:25:34 Hierarchy of psychological space
00:27:38 Spatial vagueness
00:30:57 QUALITATIVE SPATIAL REPRESENTATIONS AND REASONING (QSR)
00:34:19 Timeline of QSR history
00:37:32 Early history of QSR
00:39:26 Qualitative reasoning
00:40:52 The “start” of QSR in AI
00:41:34 “Poverty conjecture”
00:42:21 Temporal interval calculus (1D)
00:47:12 Rectangle algebra (2D)
00:49:10 Several aspects of SRR
00:50:34 Ontology of space
00:52:21 Video: USA Presidential election in 2000 depended on topology
00:56:54 Mereology
00:58:09 Mereotopology
01:01:28 Region Connection Calculus (RCC)
01:03:21 9-intersection model (9IM)
01:06:07 Mereology and topology
01:06:43 Geographical curiosity: the border between Belgium and the Netherlands
01:08:19 “SEMI-METRIC” CALCULI
01:09:22 Orientation
01:11:57 Qualitative shape descriptions
01:18:33 Mereogeometries
01:19:09 Spatio-temporal representations
01:23:15 Decidablity
01:24:13 REASONING TECHNIQUES
01:24:16 Deductive reasoning techniques
01:25:27 Inductive reasoning techniques
01:27:01 Deduction: composition tables
01:35:39 Interactive example: drawing Euler diagrams from RCC specifications
01:39:40 Spatial change
01:40:49 Conceptual neighbourhoods
01:42:22 Linguistic remark
01:44:04 Qualitative simulation of phagocytosis
01:45:02 Vagueness
01:46:50 Cognitive adequacy of QSR
01:49:54 Activity recognition in video
01:51:36 QUANTITATIVE SPATIO-TEMPORAL REASONING AND COMPUTER VISION
02:00:56 SUMMARY
Talk kindly contributed by Anthony G. Cohn in SEMF's 2022 Spacious Spatiality
semf.org.es/spatiality
TALK ABSTRACT
Agents, whether human or robotic, inhabit a spatial world, whether the real world or a virtual one; they observe it, the events that take place in it, and move around in it. They need to navigate in it, perform tasks which will affect the positions and other aspects of objects in the world. They have goals to achieve and may need to communicate about the world they inhabit with other agents, without necessarily having complete knowledge about the world. In this session, I will talk about why reasoning about space is important; what makes space “special”; the variety of spatial tasks an agent may face; and what are the particular challenges of endowing an agent with spatial intelligence. I will talk about some of the techniques the have been developed to endow agents with spatial intelligence. Amongst these are methods involving qualitative spatial reasoning which has been a particular focus of mine over the years. The real challenge is how to enable an agent to make commonsense spatial inferences, and what is the role of natural language in relation to commonsense spatial reasoning?
TALK MATERIALS
· A Survey of Qualitative Spatial Representations: eprints.whiterose.ac.uk/83881
· Automated Commonsense Spatial Reasoning: Still a Huge Challenge: eprints.whiterose.ac.uk/172683
ANTHONY G. COHN
School of Computing, Leeds University: eps.leeds.ac.uk/computing
Leeds University profile: eps.leeds.ac.uk/computing/staff/76/professor-anthony-tony-g-cohn-freng-ceng-citp
ResearchGate:researchgate.net/profile/Anthony-Cohn-2
Google Scholar: scholar.google.com/citations?user=tal4mMkAAAAJ
LinkedIn: linkedin.com/in/tonycohn
SEMF NETWORKS
Website: semf.org.es
Twitter: twitter.com/semf_nexus
LinkedIn: linkedin.com/company/semf-nexus
Instagram: instagram.com/semf.nexus
Facebook: facebook.com/semf.nexus
MUSIC
youtube.com/user/Baroquenoise
00:00:12 Introduction
00:00:56 Anthony’s background
00:03:01 SPATIAL DATA AND KNOWLEDGE
00:03:02 Ubiquity of spatial data across disciplines
00:04:10 Spatial intelligence in animals
00:05:38 Overview of the talk
00:06:10 What’s special about spatial?
00:08:47 The spatial problem solving situation
00:11:51 Qualitative spatial knowledge
00:15:10 Spatial language
00:18:48 Acquiring spatial data and knowledge
00:22:29 Spatial tasks
00:25:34 Hierarchy of psychological space
00:27:38 Spatial vagueness
00:30:57 QUALITATIVE SPATIAL REPRESENTATIONS AND REASONING (QSR)
00:34:19 Timeline of QSR history
00:37:32 Early history of QSR
00:39:26 Qualitative reasoning
00:40:52 The “start” of QSR in AI
00:41:34 “Poverty conjecture”
00:42:21 Temporal interval calculus (1D)
00:47:12 Rectangle algebra (2D)
00:49:10 Several aspects of SRR
00:50:34 Ontology of space
00:52:21 Video: USA Presidential election in 2000 depended on topology
00:56:54 Mereology
00:58:09 Mereotopology
01:01:28 Region Connection Calculus (RCC)
01:03:21 9-intersection model (9IM)
01:06:07 Mereology and topology
01:06:43 Geographical curiosity: the border between Belgium and the Netherlands
01:08:19 “SEMI-METRIC” CALCULI
01:09:22 Orientation
01:11:57 Qualitative shape descriptions
01:18:33 Mereogeometries
01:19:09 Spatio-temporal representations
01:23:15 Decidablity
01:24:13 REASONING TECHNIQUES
01:24:16 Deductive reasoning techniques
01:25:27 Inductive reasoning techniques
01:27:01 Deduction: composition tables
01:35:39 Interactive example: drawing Euler diagrams from RCC specifications
01:39:40 Spatial change
01:40:49 Conceptual neighbourhoods
01:42:22 Linguistic remark
01:44:04 Qualitative simulation of phagocytosis
01:45:02 Vagueness
01:46:50 Cognitive adequacy of QSR
01:49:54 Activity recognition in video
01:51:36 QUANTITATIVE SPATIO-TEMPORAL REASONING AND COMPUTER VISION
02:00:56 SUMMARY