Mobile intention recognition

Mobile Intention Recognition addresses problems of practical relevance for mobile system engineers: how can we make mobile assistance systems more intelligent? How can we model and recognize patterns of human behavior which span more than a limited spatial context? This text provides an overview on plan and intention recognition, ranging from the late 1970s to very recent approaches. This overview is unique as it discusses approaches with respect to the specificities of mobile intention recognition. This book covers problems from research on mobile assistance systems using methods from artificial intelligence and natural language processing. It thus addresses an extraordinary interdisciplinary audience. Cover......Page 1 Mobile Intention Recognition......Page 4 Foreword......Page 8 Acknowledgements......Page 10 Contents......Page 12 Acronyms......Page 16 1.1.1 Location-Based Services......Page 18 1.1.2.1 The on-enter/on-leave paradigm......Page 20 1.1.2.2 Spatial context ambiguity: the room-crossing problem......Page 21 1.1.2.3 Affordance ambiguity......Page 22 1.1.3 Non-Local Dependencies: The Visit/Revisit Problem......Page 23 1.2 Contributions......Page 24 1.3 Outline......Page 25 2.1.1 The Context-Model of IAMSs......Page 27 2.1.2 Crossing the Semantic Gap......Page 30 2.1.2.1 Trajectory and motion track......Page 31 2.1.2.2 Data refinement......Page 32 2.1.2.3 Segmentation......Page 34 2.1.2.4 Determining interesting behavior......Page 35 2.1.2.5 Feature extraction......Page 36 2.1.2.6 Behavior classification......Page 37 2.1.2.7 Activity recognition and classification......Page 39 2.2.1.1 Rational agents......Page 40 2.2.1.2 What are intentions?......Page 42 2.2.2 Intention Recognition and Plan Recognition......Page 44 2.3.1.1 Temporal information in mobile intention recognition......Page 47 2.3.1.2 Spatial information in mobile intention recognition......Page 49 2.3.2 Mobile Intention Recognition: A Definition......Page 54 2.3.2.2 Mobile intention recognition algorithms......Page 55 2.3.3.1 From partitioning to composition: the influence of ordering......Page 57 2.3.3.2 Spatial composition restrictions: the room-crossing problem revisited......Page 60 2.3.3.3 Adding top-down knowledge: plan libraries and behavior sequence composition......Page 62 2.3.4.1 Examples......Page 63 2.3.4.2 Nesting and crossing dependencies......Page 66 2.4 Summary and Requirements for Mobile Intention Recognition......Page 68 3.1 Finite State Machines......Page 70 3.2.1 Hypothesize and Revise......Page 72 3.2.2 Event Hierarchies......Page 73 3.3.1.1 Bayes networks and Dempster-Shafer......Page 74 3.3.1.2 Some general remarks on reasoning under uncertainty in mobile intention recognition......Page 76 3.3.2.1 The basic idea......Page 77 3.3.2.2 DBNs for mobile intention/plan recognition......Page 78 3.4.1.1 Intentional systems: a context-free grammar for representing hierarchical plans......Page 80 3.4.1.3 A note on Vilain (1990) and Sidner (1985)......Page 82 3.4.2.1 The idea of spatial disambiguation......Page 84 3.4.2.2 SGIS for mobile intention recognition......Page 85 3.4.3 Towards More Context-Sensitivity......Page 87 3.5.1 Probabilistic State-Dependent Grammars......Page 88 3.5.2 Plan Tree Grammars and Pending Sets......Page 89 3.6 Summary......Page 92 4.1.1 A State Chart Parser for SGISs......Page 94 4.1.1.2 Turning the recognizer into an incremental parser......Page 95 4.1.2 Spatial Constraint Resolution......Page 98 4.1.2.1 Scan......Page 100 4.1.2.2 Complete......Page 101 4.1.2.4 Conclusions on the runtime of the SGIS parser......Page 102 4.2.1 Spatially-Constrained Context-Free Grammars......Page 103 4.2.2 A State Chart Parser for SCCFGs......Page 105 4.2.2.1 Complete......Page 107 4.3.1 Towards More Context-Sensitivity: A Parallel to NLP......Page 108 4.3.2.1 Tree-Adjoining Grammars......Page 112 4.3.2.2 Adding Spatial Constraints to TAGs......Page 114 4.3.3 SCTAGs and the Mobile Intention Recognition Problem......Page 116 4.3.4.1 The original TAG recognizer (Schabes and Joshi, 1988)......Page 119 4.3.4.2 Turning the recognizer into an incremental parser.......Page 128 4.3.4.3 An incremental parser for SCTAG......Page 129 4.3.4.5 Conclusions on the runtime of the SCTAG parser......Page 136 4.4 Beyond Partonomies......Page 139 4.5 Summary......Page 140 5.1 The Clothes Shop Example......Page 142 5.2.1 IS and SGIS......Page 144 5.2.3 SCTAG......Page 145 5.3.1 Ambiguity Reduction with SGISs......Page 146 5.3.2 Ambiguity Reduction with SCCFGs......Page 149 5.3.3 Avoiding Inconsistency with SCTAGs......Page 150 5.4 INTENSIVE: A Simulation and Testing Environment for the IAMS Framework1......Page 152 6.1 Mobile Intention Recognition......Page 160 6.2 Spatially Constrained Grammars......Page 161 6.3.1 Temporal Constraints......Page 164 6.3.3 Probabilistic Spatially Constrained Grammars......Page 166 6.3.4 Grammar Acquisition and Behavioral Corpora......Page 167 6.3.5 Multi-Agent Intention Recognition......Page 168 References......Page 170