SlideShare une entreprise Scribd logo
1  sur  27
Rule-based Real-Time Activity Recognition
in a Smart Home Environment
Przemyslaw Woznowski
Grigoris Antoniou
10th International Web Rule Symposium (RuleML) 2016, Stony Brook, New York, USA
George Baryannis
Outline
2
Introduction
• Activity Recognition and the Internet of Things
• The SPHERE Project
• Related Work
Rule-based ADL Recognition
• Offline Version
• Online Version
• Experimental Evaluation
Conclusions & Future Work
Activity Recognition and
the Internet of Things
• Sensors have become cheaper, small, widely available
• Interconnected within an Internet of Things (IoT)
setting, benefitting from
– Distribution of resources
– Support for common naming schemas and ontologies
– Common access strategies
– Availability of computational resources
• Automated Activity Recognition (AR) requires a fusion
of multiple sensor-related low-level events
• Challenge: to locate and fuse the right pieces of
information from an IoT instance (e.g. sensor network) in
order to realise AR at the best quality of information
possible
3
Approaches for sensor-based AR
• Data-driven – exploiting machine learning techniques
Noise and uncertainty are handled well
Require large, annotated training datasets
Data conflicts are not handled well
• Knowledge-driven – leveraging logical modelling and
reasoning
No training data needed
Not as robust against noise and uncertainty
Require carefully crafted rules
4
Activity Recognition in a multi-modal
smart home environment
• Focuses on the so-called Activities of Daily Living
(ADL), with the purpose of supporting Ambient
Assisting Living (AAL) efforts
– Long-term monitoring of health-related features
– Direct assistance
• Main requirements
– Increased need for robustness against noise (due to multiple
sensors)
– Support for complex, uncertain and non-sequential scenarios
– Support for user localization within the smart home, with
minimal user involvement
– Inference of real-time, continuous streams of meaningful and
actionable events
– Less reliance on training data, since they are difficult to acquire
due to them being environment-dependent 5
The SPHERE Project
Woznowski et al. (2015)
6
• SPHERE: a
Sensor Platform
for Healthcare in a
Residential
Environment
– Common platform
of non-medical
networked sensors
– Deployed on a
home environment
testbed, the
SPHERE house
– Impact a range of
healthcare needs
simultaneously
• Chen et al.: equivalence and subsumption reasoning on
ontologies modelling both sensors and activities
Both offline and real-time modes, incrementally-specific
recognition
Requires activities to be performed in a predefined, strictly
sequential order and fixed time intervals
• MetaQ: SPARQL-based reasoning on sensor data
represented as RDF graphs
Recognition building from atomic gestures to complex activities
Works only offline, does not take into account missing activities
Related Work (1)
7
• Skarlatidis et al.: hybrid approach, combining event
calculus reasoning with Markov Logic Networks
High recognition rates, robustness against missing data
Only focuses on posture and movement-related activities, as
opposed to complex ADL scenarios
• Helaoui et al.: hybrid approach, employing a probabilistic
DL reasoner
Recognition building from atomic gestures to complex activities
Requires training data, works only offline, no support for
temporal features
Related Work (2)
8
Outline
9
Introduction
• Activity Recognition
• The SPHERE Project
• Related Work
Rule-based ADL Recognition
• Offline Version
• Online Version
• Experimental Evaluation
Conclusions & Future Work
10
• Rule base
– rules defined by examining collected
sensor data from scripted experiments
• Fact base
– derived from sensor data
• The system operates in two modes
– Offline: precollected sensor data are stored as individual facts
• Can provide activity reports for past periods (e.g. hourly or daily)
– Real-time: facts represent each deployed sensor node and store
its current state/value (as well as its previous one)
• Recognises activities as soon as the associated sensor events
happen
Rule
Base
Fact
Base
Inference Engine
(JESS)
“Expert” knowledge Sensor data
Rule-based System Overview
11
• Environmental Sensors
– Door contact, electricity meters, water flow meters, PIR
– Ambient light useful only when the effect of sunlight is minimal
(i.e. the sun is below the horizon)
– Scripted experiment data do not yield patterns from ambient
noise, dust, humidity and temperature
• Video Sensors
– 2D bounding box coordinates
– Depth coordinates of 3D bounding box
Fact Base
12
PIR-based
Location
Camera-
based
Location
Fused
Location
Second Level Rules
Higher Level Rules
Door
Interaction
Electrical
Devices
Water
Flow
Water
Flow
Clean-up
Atomic Activity Rules
Rule Base
rules assume single inhabitant scenarios
13
• Detect changes in sensor values within their reporting
windows
– From >0 to 0: OpenDoor / SwitchOff
– From 0 to >0: CloseDoor / SwitchOn
Doors and Electrical Devices
14
• Water meters do not have a reporting period, only report
instantaneously
– Positive flow value: OpenTap
– Zero flow value: CloseTap
• “Clean up” rules follow to keep only the earliest events
for each distinct opening or closing occurrence
– If there is no close tap activity between two consecutive open tap
activities, remove the latest one
Water Flow
15
• Rules so far recognise atomic activities
• Higher-level rules progressively combine recognised
activities to infer activities of increasingly higher
complexity
– SwitchOn(device,t1) and SwitchOff(device,t2)  Use(device, t1,
t2)
– SwitchOn(tv,t1) and SwitchOff(tv,t2)  WatchingTV(t1, t2)
– Use of taps in kitchen or bathroom  WashHands or
WashFace
– Use of taps in bathroom  BrushTeeth or Bathing/Showering
– Use kettle and close tap in kitchen  PreparingDrink
– Open fridge and use toaster  PreparingSnack
– PreparingDrink or PreparingSnack and use of taps in kitchen 
WashDishes
Complex Activities
16
• Basic PIR rule places user in a specific room, from the
time PIR is activated till it’s deactivated
– Sequences in the same room merged if temporally close or user
not in a different room in between
• Basic video rule places user in a specific room, for as
long as the associated camera reports bounding box
coordinates
• Detect ghost sequences since they severely
compromise validity
– Length of less than 30 frames
– Stuck in the same coordinates for more than 30 frames
– Width and/or height of box consistently and unjustifiably small,
in correlation with depth
Localisation Rules
17
• Assign confidence values to PIR and camera location reports
– PIR: confidence inversely proportional to the number of PIR sensors
simultaneously reporting motion
– Video: confidence depends on the probability of being a ghost,
based on the detection heuristics
• If only a single source reports a location, it is assumed to
hold (with the associated confidence value)
• If PIR and video report the same location, it is assumed to
hold (with confidence values summed)
• If PIR and video disagree, the correct location is the one
associated with a recognised atomic activity
• If both or neither disagreeing reports are supported by an
activity, we assume the one with the higher confidence
holds
– If confidences are equal, we trust PIR
Fused Location
18
• Facts now represent the state of each distinct sensor
– Instead of the history of sensor events
– To detect state change, previous state is also stored
• Changes are necessary only for rules at the lowest level
– Second and higher-level rules remain unchanged
• Transparent to the way sensor events are generated
• Any state change event is linked to a related atomic
activity
– Holds for DC sensors, electricity and water flow meters
– Rules fire only once when sensor values change – no need for
“clean up” rules
From Offline to Real-time
19
• Each consecutive activation/deactivation of a PIR sensor
corresponds to the user being in the associated room
• Subsequent activations extend the user’s stay when
– Activation directly follows the last deactivation
– The elapsed time between them does not exceed a threshold
– No activation has taken place in a different room in the
meantime
• State-based approach is not applicable for video sensors
– Video cameras do not broadcast a single value
• Each reported bounding box is stored briefly
– Combined to create facts that represent a period of stay in a
room
– The same heuristics used for ghost detection
Online Localisation (1)
20
• Each time a PIR sensor is activated, the system fuses
available information to decide on its validity
– If there is no active video sequence and no activity detected, we
assume PIR is valid
– If the active video sequence with the highest confidence agrees
with PIR, we conclude the user is in the room, summing
confidence values
– If video reports a different room, we assume the user is in the
room where the most recently recognised atomic activity was
performed
Online Localisation (2): Fusion
21
• Offline and real-time versions implemented in Java,
using Jess as a rule engine
– Implemented rules designed to accommodate variable reporting
periods
• Real-time version built as an MQTT client
– Sensor messages are broadcast in separate threads
• 10 participants executed an ADL script of half-hour
duration, twice, in the SPHERE house
– Ground-truth data acquired through annotation of video images
collected using a head-mounted camera
– Subset of performed activities that are recognised: interaction
with doors, electrical devices and water taps, preparing a
snack/drink, washing hands/dishes, brushing teeth,
bathing/showering
Implementation and Data Collection
22
• TP (true positive): activity performed and recognised
• FP (false positive): activity not performed but recognised
• FN (false negative): activity performed but not recognised
• 𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =
𝑇𝑃
𝑇𝑃+𝐹𝑃
%, 𝑟𝑒𝑐𝑎𝑙𝑙 =
𝑇𝑃
𝑇𝑃+𝐹𝑁
%
Evaluation Results
Outline
23
Introduction
• Activity Recognition
• The SPHERE Project
• Related Work
Rule-based ADL Recognition
• Offline Version
• Online Version
• Experimental Evaluation
Conclusions & Future Work
Concluding remarks
• Rule-based system capable of operating on both
historical and real-time, multi-modal sensor data
acquired in a smart home
– Bottom-up, multi-level rules to support complex ADL scenarios
– Non-deterministic patterns to account for missing activities
• Sensor fusion and heuristics to achieve robustness
against noise
– 95% recall and 88% precision on average for a significant subset
of activities
– 93% room-level localisation accuracy due to effective ghost
detection and location fusion rules
24
25
• Integrate wearable sensor data
– Infer activities unidentifiable with only the other sensors
– Improve localisation accuracy or provide an alternative to video
cameras when they are not available/allowed
• Explore multi-inhabitant scenarios
– Use localisation results to pin down activities to the person
performing them
– For some activities, localisation needs to be more fine-grained
than room-level
• Explore hybrid approach with Machine Learning
research within SPHERE
– Incorporate rules as features in ML algorithms
– Use rules that act on the results of ML algorithms
– Devise ML techniques to learn rules
Current and Future Work
26
Thank you! Questions?
g.bargiannis@hud.ac.uk
http://www.irc-sphere.ac.uk
27
• Chen, L., Nugent, C.D., Wang, H.: A knowledge-driven approach to activity
recognition in smart homes. IEEE Trans. Knowl. Data Eng. 24(6), 961–974
(2012)
• Filippaki, C., Antoniou, G., Tsamardinos, I.: Using constraint optimization for
conflict resolution and detail control in activity recognition. In: Keyson, D.V.,
Maher, M.L., Streitz, N., Cheok, A., Augusto, J.C., Wichert, R., Englebienne, G.,
Aghajan, H., Krose, B.J.A. (eds.) AmI 2011. LNCS, vol. 7040, pp. 51–60.
Springer, Heidelberg (2011)
• Helaoui, R., Riboni, D., Stuckenschmidt, H.: A probabilistic ontological
framework for the recognition of multilevel human activities. In: Mattern, F.,
Santini, S., Canny, J.F., Langheinrich, M., Rekimoto, J. (eds.) UbiComp 2013,
pp. 345–354. ACM (2013)
• Meditskos, G., Dasiopoulou, S., Kompatsiaris, I.: MetaQ: a knowledge-driven
framework for context-aware activity recognition combining SPARQL and OWL
2 activity patterns. Pervasive Mob. Comput. 25, 104–124 (2016)
• Skarlatidis, A., Paliouras, G., Artikis, A., Vouros, G.A.: Probabilistic event
calculus for event recognition. ACM Trans. Comput. Log. 16(2), 11:1–11:37
(2015)
• Woznowski, P., Fafoutis, X., Song, T., Hannuna, S., Camplani, M., Tao, L.,
Paiement, A., Mellios, E., Haghighi, M., Zhu, N., et al.: A multi-modal sensor
infrastructure for healthcare in a residential environment. In: 2015 IEEE
International Conference on Communication Workshop, pp. 271–277. IEEE
(2015)
References

Contenu connexe

Tendances

Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...
Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...
Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...Mysa Vijay
 
Forest fire detection & alarm thermal imaging camera
Forest fire detection & alarm thermal imaging camera  Forest fire detection & alarm thermal imaging camera
Forest fire detection & alarm thermal imaging camera Elsa Wang
 
The cougar approach to in-network query processing in sensor networks
The cougar approach to in-network query processing in sensor networksThe cougar approach to in-network query processing in sensor networks
The cougar approach to in-network query processing in sensor networksDilini Muthumala
 
Interoperable data management OBSEA
Interoperable data management OBSEAInteroperable data management OBSEA
Interoperable data management OBSEAÆdel Aerospace GmbH
 
XGSN: An Open-source Semantic Sensing Middleware for the Web of Things
XGSN: An Open-source Semantic Sensing Middleware for the Web of ThingsXGSN: An Open-source Semantic Sensing Middleware for the Web of Things
XGSN: An Open-source Semantic Sensing Middleware for the Web of ThingsJean-Paul Calbimonte
 
Formatted Paper_References added
Formatted Paper_References addedFormatted Paper_References added
Formatted Paper_References addedMark Kurtz
 
Smart classroom using arduino with internet of thing(io t)
Smart classroom using arduino with internet of thing(io t)Smart classroom using arduino with internet of thing(io t)
Smart classroom using arduino with internet of thing(io t)ainaa aa
 
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...IRJET Journal
 
GSN Global Sensor Networks for Environmental Data Management
GSN Global Sensor Networks for Environmental Data ManagementGSN Global Sensor Networks for Environmental Data Management
GSN Global Sensor Networks for Environmental Data ManagementJean-Paul Calbimonte
 
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...IRJET Journal
 
Location Based Development Using Xamarin
Location Based Development Using XamarinLocation Based Development Using Xamarin
Location Based Development Using XamarinKym Phillpotts
 

Tendances (13)

Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...
Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...
Mobile and Web Applications for Sensing Hazardous Room Temperature using Wire...
 
Forest fire detection & alarm thermal imaging camera
Forest fire detection & alarm thermal imaging camera  Forest fire detection & alarm thermal imaging camera
Forest fire detection & alarm thermal imaging camera
 
The cougar approach to in-network query processing in sensor networks
The cougar approach to in-network query processing in sensor networksThe cougar approach to in-network query processing in sensor networks
The cougar approach to in-network query processing in sensor networks
 
Interoperable data management OBSEA
Interoperable data management OBSEAInteroperable data management OBSEA
Interoperable data management OBSEA
 
XGSN: An Open-source Semantic Sensing Middleware for the Web of Things
XGSN: An Open-source Semantic Sensing Middleware for the Web of ThingsXGSN: An Open-source Semantic Sensing Middleware for the Web of Things
XGSN: An Open-source Semantic Sensing Middleware for the Web of Things
 
Formatted Paper_References added
Formatted Paper_References addedFormatted Paper_References added
Formatted Paper_References added
 
Smart classroom using arduino with internet of thing(io t)
Smart classroom using arduino with internet of thing(io t)Smart classroom using arduino with internet of thing(io t)
Smart classroom using arduino with internet of thing(io t)
 
IoT Introduction
IoT Introduction IoT Introduction
IoT Introduction
 
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...
IRJET-An Automatic Fire Detection and Warning System under Home Video Surveil...
 
GSN Global Sensor Networks for Environmental Data Management
GSN Global Sensor Networks for Environmental Data ManagementGSN Global Sensor Networks for Environmental Data Management
GSN Global Sensor Networks for Environmental Data Management
 
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...
IRJET- An IoT Based Forest Fire Detection and Prevention System using Raspber...
 
Location Based Development Using Xamarin
Location Based Development Using XamarinLocation Based Development Using Xamarin
Location Based Development Using Xamarin
 
X-GSN in OpenIoT SummerSchool
X-GSN in OpenIoT SummerSchoolX-GSN in OpenIoT SummerSchool
X-GSN in OpenIoT SummerSchool
 

En vedette

Semantic human activity detection in videos
Semantic human activity detection in videosSemantic human activity detection in videos
Semantic human activity detection in videosHirantha Pradeep
 
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon Transform
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon TransformHuman Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon Transform
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon TransformFadwa Fouad
 
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...PyData
 
マーケティングサイエンス徹底入門と実践Part2
マーケティングサイエンス徹底入門と実践Part2マーケティングサイエンス徹底入門と実践Part2
マーケティングサイエンス徹底入門と実践Part2宏喜 佐野
 
10分で分かるr言語入門ver2.7
10分で分かるr言語入門ver2.710分で分かるr言語入門ver2.7
10分で分かるr言語入門ver2.7Nobuaki Oshiro
 
10分で分かるr言語入門ver2.15 15 1010
10分で分かるr言語入門ver2.15 15 101010分で分かるr言語入門ver2.15 15 1010
10分で分かるr言語入門ver2.15 15 1010Nobuaki Oshiro
 
Rで学ぶ離散選択モデル
Rで学ぶ離散選択モデルRで学ぶ離散選択モデル
Rで学ぶ離散選択モデル宏喜 佐野
 
20161127 doradora09 japanr2016_lt
20161127 doradora09 japanr2016_lt20161127 doradora09 japanr2016_lt
20161127 doradora09 japanr2016_ltNobuaki Oshiro
 
Inria - White Paper - Artificial intelligence, current challenges and Inria's...
Inria - White Paper - Artificial intelligence, current challenges and Inria's...Inria - White Paper - Artificial intelligence, current challenges and Inria's...
Inria - White Paper - Artificial intelligence, current challenges and Inria's...Inria
 
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリングベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング宏喜 佐野
 
Rで学ぶデータマイニングI 第8章〜第13章
Rで学ぶデータマイニングI 第8章〜第13章Rで学ぶデータマイニングI 第8章〜第13章
Rで学ぶデータマイニングI 第8章〜第13章Prunus 1350
 

En vedette (11)

Semantic human activity detection in videos
Semantic human activity detection in videosSemantic human activity detection in videos
Semantic human activity detection in videos
 
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon Transform
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon TransformHuman Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon Transform
Human Action Recognition in Videos Employing 2DPCA on 2DHOOF and Radon Transform
 
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...
Introduction to Action Recognition in Python by Bertrand Nouvel, Jonathan Kel...
 
マーケティングサイエンス徹底入門と実践Part2
マーケティングサイエンス徹底入門と実践Part2マーケティングサイエンス徹底入門と実践Part2
マーケティングサイエンス徹底入門と実践Part2
 
10分で分かるr言語入門ver2.7
10分で分かるr言語入門ver2.710分で分かるr言語入門ver2.7
10分で分かるr言語入門ver2.7
 
10分で分かるr言語入門ver2.15 15 1010
10分で分かるr言語入門ver2.15 15 101010分で分かるr言語入門ver2.15 15 1010
10分で分かるr言語入門ver2.15 15 1010
 
Rで学ぶ離散選択モデル
Rで学ぶ離散選択モデルRで学ぶ離散選択モデル
Rで学ぶ離散選択モデル
 
20161127 doradora09 japanr2016_lt
20161127 doradora09 japanr2016_lt20161127 doradora09 japanr2016_lt
20161127 doradora09 japanr2016_lt
 
Inria - White Paper - Artificial intelligence, current challenges and Inria's...
Inria - White Paper - Artificial intelligence, current challenges and Inria's...Inria - White Paper - Artificial intelligence, current challenges and Inria's...
Inria - White Paper - Artificial intelligence, current challenges and Inria's...
 
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリングベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング
ベイジアンモデリングによるマーケティングサイエンス〜状態空間モデルを用いたモデリング
 
Rで学ぶデータマイニングI 第8章〜第13章
Rで学ぶデータマイニングI 第8章〜第13章Rで学ぶデータマイニングI 第8章〜第13章
Rで学ぶデータマイニングI 第8章〜第13章
 

Similaire à Rule-based Real-Time Activity Recognition in a Smart Home Environment

Rule based expert system
Rule based expert systemRule based expert system
Rule based expert systemAbhishek Kori
 
Data, Big Data and real time analytics for Connected Devices
Data, Big Data and real time analytics for Connected DevicesData, Big Data and real time analytics for Connected Devices
Data, Big Data and real time analytics for Connected DevicesSrinath Perera
 
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...Diego López-de-Ipiña González-de-Artaza
 
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...Edge AI and Vision Alliance
 
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...Eswar Publications
 
IoT Levels and Deployment Templates
IoT Levels and Deployment TemplatesIoT Levels and Deployment Templates
IoT Levels and Deployment TemplatesPrakash Honnur
 
Lect14-15 iotlevelsanddeploymenttemplates.pdf
Lect14-15 iotlevelsanddeploymenttemplates.pdfLect14-15 iotlevelsanddeploymenttemplates.pdf
Lect14-15 iotlevelsanddeploymenttemplates.pdfjayakumar703719
 
A benchmark dataset to evaluate sensor displacement in activity recognition
A benchmark dataset to evaluate sensor displacement in activity recognitionA benchmark dataset to evaluate sensor displacement in activity recognition
A benchmark dataset to evaluate sensor displacement in activity recognitionOresti Banos
 
III CSE IoT Unit - I.pptx
III CSE IoT Unit - I.pptxIII CSE IoT Unit - I.pptx
III CSE IoT Unit - I.pptxAvinashAvuthu2
 
From Virtual to Real World: Applying Animation to Design the Activity Recogni...
From Virtual to Real World: Applying Animation to Design the Activity Recogni...From Virtual to Real World: Applying Animation to Design the Activity Recogni...
From Virtual to Real World: Applying Animation to Design the Activity Recogni...sugiuralab
 
Collecting big data in cinemas to improve recommendation systems - a model wi...
Collecting big data in cinemas to improve recommendation systems - a model wi...Collecting big data in cinemas to improve recommendation systems - a model wi...
Collecting big data in cinemas to improve recommendation systems - a model wi...ICDEcCnferenece
 
4th Year Project Presentation Slides
4th Year Project Presentation Slides4th Year Project Presentation Slides
4th Year Project Presentation SlidesItrat Rahman
 
Intelligent Wireless Sensor Network Simulation
Intelligent Wireless Sensor Network SimulationIntelligent Wireless Sensor Network Simulation
Intelligent Wireless Sensor Network Simulationcatherine roussey
 
Real time visualization of structured things
Real time visualization of structured thingsReal time visualization of structured things
Real time visualization of structured thingsNurul Amin Choudhury
 
Sensor Networks Introduction and Architecture
Sensor Networks Introduction and ArchitectureSensor Networks Introduction and Architecture
Sensor Networks Introduction and ArchitecturePeriyanayagiS
 
Iit kgp workshop
Iit kgp workshopIit kgp workshop
Iit kgp workshopArpan Pal
 
IRJET - Prisoners Surveillance System
IRJET - Prisoners Surveillance SystemIRJET - Prisoners Surveillance System
IRJET - Prisoners Surveillance SystemIRJET Journal
 
Artificial Intelligence for Automated Decision Support Project
Artificial Intelligence for Automated Decision Support ProjectArtificial Intelligence for Automated Decision Support Project
Artificial Intelligence for Automated Decision Support ProjectValerii Klymchuk
 
Cps innovation lab kolkata iiest
Cps innovation lab kolkata iiestCps innovation lab kolkata iiest
Cps innovation lab kolkata iiestArpan Pal
 
Composite Intrusion Detection in Process Control Networks
Composite Intrusion Detection in Process Control NetworksComposite Intrusion Detection in Process Control Networks
Composite Intrusion Detection in Process Control Networksguest8fdee6
 

Similaire à Rule-based Real-Time Activity Recognition in a Smart Home Environment (20)

Rule based expert system
Rule based expert systemRule based expert system
Rule based expert system
 
Data, Big Data and real time analytics for Connected Devices
Data, Big Data and real time analytics for Connected DevicesData, Big Data and real time analytics for Connected Devices
Data, Big Data and real time analytics for Connected Devices
 
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...
The quest for Ubiquitous Computing: from Ambient Intelligence to the combinat...
 
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...
“Making Sense of Sensors: Combining Visual, Laser and Wireless Sensors to Pow...
 
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...
Nuzzer algorithm based Human Tracking and Security System for Device-Free Pas...
 
IoT Levels and Deployment Templates
IoT Levels and Deployment TemplatesIoT Levels and Deployment Templates
IoT Levels and Deployment Templates
 
Lect14-15 iotlevelsanddeploymenttemplates.pdf
Lect14-15 iotlevelsanddeploymenttemplates.pdfLect14-15 iotlevelsanddeploymenttemplates.pdf
Lect14-15 iotlevelsanddeploymenttemplates.pdf
 
A benchmark dataset to evaluate sensor displacement in activity recognition
A benchmark dataset to evaluate sensor displacement in activity recognitionA benchmark dataset to evaluate sensor displacement in activity recognition
A benchmark dataset to evaluate sensor displacement in activity recognition
 
III CSE IoT Unit - I.pptx
III CSE IoT Unit - I.pptxIII CSE IoT Unit - I.pptx
III CSE IoT Unit - I.pptx
 
From Virtual to Real World: Applying Animation to Design the Activity Recogni...
From Virtual to Real World: Applying Animation to Design the Activity Recogni...From Virtual to Real World: Applying Animation to Design the Activity Recogni...
From Virtual to Real World: Applying Animation to Design the Activity Recogni...
 
Collecting big data in cinemas to improve recommendation systems - a model wi...
Collecting big data in cinemas to improve recommendation systems - a model wi...Collecting big data in cinemas to improve recommendation systems - a model wi...
Collecting big data in cinemas to improve recommendation systems - a model wi...
 
4th Year Project Presentation Slides
4th Year Project Presentation Slides4th Year Project Presentation Slides
4th Year Project Presentation Slides
 
Intelligent Wireless Sensor Network Simulation
Intelligent Wireless Sensor Network SimulationIntelligent Wireless Sensor Network Simulation
Intelligent Wireless Sensor Network Simulation
 
Real time visualization of structured things
Real time visualization of structured thingsReal time visualization of structured things
Real time visualization of structured things
 
Sensor Networks Introduction and Architecture
Sensor Networks Introduction and ArchitectureSensor Networks Introduction and Architecture
Sensor Networks Introduction and Architecture
 
Iit kgp workshop
Iit kgp workshopIit kgp workshop
Iit kgp workshop
 
IRJET - Prisoners Surveillance System
IRJET - Prisoners Surveillance SystemIRJET - Prisoners Surveillance System
IRJET - Prisoners Surveillance System
 
Artificial Intelligence for Automated Decision Support Project
Artificial Intelligence for Automated Decision Support ProjectArtificial Intelligence for Automated Decision Support Project
Artificial Intelligence for Automated Decision Support Project
 
Cps innovation lab kolkata iiest
Cps innovation lab kolkata iiestCps innovation lab kolkata iiest
Cps innovation lab kolkata iiest
 
Composite Intrusion Detection in Process Control Networks
Composite Intrusion Detection in Process Control NetworksComposite Intrusion Detection in Process Control Networks
Composite Intrusion Detection in Process Control Networks
 

Plus de George Baryannis

A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...
A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...
A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...George Baryannis
 
A Novel Specification and Composition Language for Services
A Novel Specification and Composition Language for ServicesA Novel Specification and Composition Language for Services
A Novel Specification and Composition Language for ServicesGeorge Baryannis
 
Fluent Calculus-based Semantic Web Service Composition and Verification using...
Fluent Calculus-based Semantic Web Service Composition and Verification using...Fluent Calculus-based Semantic Web Service Composition and Verification using...
Fluent Calculus-based Semantic Web Service Composition and Verification using...George Baryannis
 
WSSL: A Fluent Calculus-based Language for Web Service Specifications
WSSL: A Fluent Calculus-based Language for Web Service SpecificationsWSSL: A Fluent Calculus-based Language for Web Service Specifications
WSSL: A Fluent Calculus-based Language for Web Service SpecificationsGeorge Baryannis
 
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...George Baryannis
 
Towards Realizing Dynamic QoS-aware Web Service Composition
Towards Realizing  Dynamic QoS-aware Web Service CompositionTowards Realizing  Dynamic QoS-aware Web Service Composition
Towards Realizing Dynamic QoS-aware Web Service CompositionGeorge Baryannis
 
Deriving Specifications for Composite Web Services
Deriving Specifications for Composite Web ServicesDeriving Specifications for Composite Web Services
Deriving Specifications for Composite Web ServicesGeorge Baryannis
 
Deriving composite service specifications
Deriving composite service specificationsDeriving composite service specifications
Deriving composite service specificationsGeorge Baryannis
 

Plus de George Baryannis (8)

A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...
A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...
A Trajectory Calculus for Qualitative Spatial Reasoning Using Answer Set Prog...
 
A Novel Specification and Composition Language for Services
A Novel Specification and Composition Language for ServicesA Novel Specification and Composition Language for Services
A Novel Specification and Composition Language for Services
 
Fluent Calculus-based Semantic Web Service Composition and Verification using...
Fluent Calculus-based Semantic Web Service Composition and Verification using...Fluent Calculus-based Semantic Web Service Composition and Verification using...
Fluent Calculus-based Semantic Web Service Composition and Verification using...
 
WSSL: A Fluent Calculus-based Language for Web Service Specifications
WSSL: A Fluent Calculus-based Language for Web Service SpecificationsWSSL: A Fluent Calculus-based Language for Web Service Specifications
WSSL: A Fluent Calculus-based Language for Web Service Specifications
 
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...
Lifecycle Management of Service-based Applications on Multi-Clouds: A Resear...
 
Towards Realizing Dynamic QoS-aware Web Service Composition
Towards Realizing  Dynamic QoS-aware Web Service CompositionTowards Realizing  Dynamic QoS-aware Web Service Composition
Towards Realizing Dynamic QoS-aware Web Service Composition
 
Deriving Specifications for Composite Web Services
Deriving Specifications for Composite Web ServicesDeriving Specifications for Composite Web Services
Deriving Specifications for Composite Web Services
 
Deriving composite service specifications
Deriving composite service specificationsDeriving composite service specifications
Deriving composite service specifications
 

Dernier

My Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle BaileyMy Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle Baileyhlharris
 
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara Services
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara ServicesVVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara Services
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara ServicesPooja Nehwal
 
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptxChiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptxraffaeleoman
 
Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510Vipesco
 
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...Kayode Fayemi
 
Causes of poverty in France presentation.pptx
Causes of poverty in France presentation.pptxCauses of poverty in France presentation.pptx
Causes of poverty in France presentation.pptxCamilleBoulbin1
 
Uncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac FolorunsoUncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac FolorunsoKayode Fayemi
 
Report Writing Webinar Training
Report Writing Webinar TrainingReport Writing Webinar Training
Report Writing Webinar TrainingKylaCullinane
 
Air breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animalsAir breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animalsaqsarehman5055
 
lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.lodhisaajjda
 
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdfAWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdfSkillCertProExams
 
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...Delhi Call girls
 
ANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docxANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docxNikitaBankoti2
 
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort ServiceBDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort ServiceDelhi Call girls
 
SaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, YardstickSaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, Yardsticksaastr
 
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptxMohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptxmohammadalnahdi22
 
Dreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video TreatmentDreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video Treatmentnswingard
 
If this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New NigeriaIf this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New NigeriaKayode Fayemi
 
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...Hasting Chen
 

Dernier (20)

My Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle BaileyMy Presentation "In Your Hands" by Halle Bailey
My Presentation "In Your Hands" by Halle Bailey
 
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara Services
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara ServicesVVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara Services
VVIP Call Girls Nalasopara : 9892124323, Call Girls in Nalasopara Services
 
ICT role in 21st century education and it's challenges.pdf
ICT role in 21st century education and it's challenges.pdfICT role in 21st century education and it's challenges.pdf
ICT role in 21st century education and it's challenges.pdf
 
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptxChiulli_Aurora_Oman_Raffaele_Beowulf.pptx
Chiulli_Aurora_Oman_Raffaele_Beowulf.pptx
 
Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510Thirunelveli call girls Tamil escorts 7877702510
Thirunelveli call girls Tamil escorts 7877702510
 
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...
Governance and Nation-Building in Nigeria: Some Reflections on Options for Po...
 
Causes of poverty in France presentation.pptx
Causes of poverty in France presentation.pptxCauses of poverty in France presentation.pptx
Causes of poverty in France presentation.pptx
 
Uncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac FolorunsoUncommon Grace The Autobiography of Isaac Folorunso
Uncommon Grace The Autobiography of Isaac Folorunso
 
Report Writing Webinar Training
Report Writing Webinar TrainingReport Writing Webinar Training
Report Writing Webinar Training
 
Air breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animalsAir breathing and respiratory adaptations in diver animals
Air breathing and respiratory adaptations in diver animals
 
lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.lONG QUESTION ANSWER PAKISTAN STUDIES10.
lONG QUESTION ANSWER PAKISTAN STUDIES10.
 
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdfAWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
AWS Data Engineer Associate (DEA-C01) Exam Dumps 2024.pdf
 
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
Busty Desi⚡Call Girls in Sector 51 Noida Escorts >༒8448380779 Escort Service-...
 
ANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docxANCHORING SCRIPT FOR A CULTURAL EVENT.docx
ANCHORING SCRIPT FOR A CULTURAL EVENT.docx
 
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort ServiceBDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
BDSM⚡Call Girls in Sector 97 Noida Escorts >༒8448380779 Escort Service
 
SaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, YardstickSaaStr Workshop Wednesday w/ Lucas Price, Yardstick
SaaStr Workshop Wednesday w/ Lucas Price, Yardstick
 
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptxMohammad_Alnahdi_Oral_Presentation_Assignment.pptx
Mohammad_Alnahdi_Oral_Presentation_Assignment.pptx
 
Dreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video TreatmentDreaming Marissa Sánchez Music Video Treatment
Dreaming Marissa Sánchez Music Video Treatment
 
If this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New NigeriaIf this Giant Must Walk: A Manifesto for a New Nigeria
If this Giant Must Walk: A Manifesto for a New Nigeria
 
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...
Re-membering the Bard: Revisiting The Compleat Wrks of Wllm Shkspr (Abridged)...
 

Rule-based Real-Time Activity Recognition in a Smart Home Environment

  • 1. Rule-based Real-Time Activity Recognition in a Smart Home Environment Przemyslaw Woznowski Grigoris Antoniou 10th International Web Rule Symposium (RuleML) 2016, Stony Brook, New York, USA George Baryannis
  • 2. Outline 2 Introduction • Activity Recognition and the Internet of Things • The SPHERE Project • Related Work Rule-based ADL Recognition • Offline Version • Online Version • Experimental Evaluation Conclusions & Future Work
  • 3. Activity Recognition and the Internet of Things • Sensors have become cheaper, small, widely available • Interconnected within an Internet of Things (IoT) setting, benefitting from – Distribution of resources – Support for common naming schemas and ontologies – Common access strategies – Availability of computational resources • Automated Activity Recognition (AR) requires a fusion of multiple sensor-related low-level events • Challenge: to locate and fuse the right pieces of information from an IoT instance (e.g. sensor network) in order to realise AR at the best quality of information possible 3
  • 4. Approaches for sensor-based AR • Data-driven – exploiting machine learning techniques Noise and uncertainty are handled well Require large, annotated training datasets Data conflicts are not handled well • Knowledge-driven – leveraging logical modelling and reasoning No training data needed Not as robust against noise and uncertainty Require carefully crafted rules 4
  • 5. Activity Recognition in a multi-modal smart home environment • Focuses on the so-called Activities of Daily Living (ADL), with the purpose of supporting Ambient Assisting Living (AAL) efforts – Long-term monitoring of health-related features – Direct assistance • Main requirements – Increased need for robustness against noise (due to multiple sensors) – Support for complex, uncertain and non-sequential scenarios – Support for user localization within the smart home, with minimal user involvement – Inference of real-time, continuous streams of meaningful and actionable events – Less reliance on training data, since they are difficult to acquire due to them being environment-dependent 5
  • 6. The SPHERE Project Woznowski et al. (2015) 6 • SPHERE: a Sensor Platform for Healthcare in a Residential Environment – Common platform of non-medical networked sensors – Deployed on a home environment testbed, the SPHERE house – Impact a range of healthcare needs simultaneously
  • 7. • Chen et al.: equivalence and subsumption reasoning on ontologies modelling both sensors and activities Both offline and real-time modes, incrementally-specific recognition Requires activities to be performed in a predefined, strictly sequential order and fixed time intervals • MetaQ: SPARQL-based reasoning on sensor data represented as RDF graphs Recognition building from atomic gestures to complex activities Works only offline, does not take into account missing activities Related Work (1) 7
  • 8. • Skarlatidis et al.: hybrid approach, combining event calculus reasoning with Markov Logic Networks High recognition rates, robustness against missing data Only focuses on posture and movement-related activities, as opposed to complex ADL scenarios • Helaoui et al.: hybrid approach, employing a probabilistic DL reasoner Recognition building from atomic gestures to complex activities Requires training data, works only offline, no support for temporal features Related Work (2) 8
  • 9. Outline 9 Introduction • Activity Recognition • The SPHERE Project • Related Work Rule-based ADL Recognition • Offline Version • Online Version • Experimental Evaluation Conclusions & Future Work
  • 10. 10 • Rule base – rules defined by examining collected sensor data from scripted experiments • Fact base – derived from sensor data • The system operates in two modes – Offline: precollected sensor data are stored as individual facts • Can provide activity reports for past periods (e.g. hourly or daily) – Real-time: facts represent each deployed sensor node and store its current state/value (as well as its previous one) • Recognises activities as soon as the associated sensor events happen Rule Base Fact Base Inference Engine (JESS) “Expert” knowledge Sensor data Rule-based System Overview
  • 11. 11 • Environmental Sensors – Door contact, electricity meters, water flow meters, PIR – Ambient light useful only when the effect of sunlight is minimal (i.e. the sun is below the horizon) – Scripted experiment data do not yield patterns from ambient noise, dust, humidity and temperature • Video Sensors – 2D bounding box coordinates – Depth coordinates of 3D bounding box Fact Base
  • 12. 12 PIR-based Location Camera- based Location Fused Location Second Level Rules Higher Level Rules Door Interaction Electrical Devices Water Flow Water Flow Clean-up Atomic Activity Rules Rule Base rules assume single inhabitant scenarios
  • 13. 13 • Detect changes in sensor values within their reporting windows – From >0 to 0: OpenDoor / SwitchOff – From 0 to >0: CloseDoor / SwitchOn Doors and Electrical Devices
  • 14. 14 • Water meters do not have a reporting period, only report instantaneously – Positive flow value: OpenTap – Zero flow value: CloseTap • “Clean up” rules follow to keep only the earliest events for each distinct opening or closing occurrence – If there is no close tap activity between two consecutive open tap activities, remove the latest one Water Flow
  • 15. 15 • Rules so far recognise atomic activities • Higher-level rules progressively combine recognised activities to infer activities of increasingly higher complexity – SwitchOn(device,t1) and SwitchOff(device,t2)  Use(device, t1, t2) – SwitchOn(tv,t1) and SwitchOff(tv,t2)  WatchingTV(t1, t2) – Use of taps in kitchen or bathroom  WashHands or WashFace – Use of taps in bathroom  BrushTeeth or Bathing/Showering – Use kettle and close tap in kitchen  PreparingDrink – Open fridge and use toaster  PreparingSnack – PreparingDrink or PreparingSnack and use of taps in kitchen  WashDishes Complex Activities
  • 16. 16 • Basic PIR rule places user in a specific room, from the time PIR is activated till it’s deactivated – Sequences in the same room merged if temporally close or user not in a different room in between • Basic video rule places user in a specific room, for as long as the associated camera reports bounding box coordinates • Detect ghost sequences since they severely compromise validity – Length of less than 30 frames – Stuck in the same coordinates for more than 30 frames – Width and/or height of box consistently and unjustifiably small, in correlation with depth Localisation Rules
  • 17. 17 • Assign confidence values to PIR and camera location reports – PIR: confidence inversely proportional to the number of PIR sensors simultaneously reporting motion – Video: confidence depends on the probability of being a ghost, based on the detection heuristics • If only a single source reports a location, it is assumed to hold (with the associated confidence value) • If PIR and video report the same location, it is assumed to hold (with confidence values summed) • If PIR and video disagree, the correct location is the one associated with a recognised atomic activity • If both or neither disagreeing reports are supported by an activity, we assume the one with the higher confidence holds – If confidences are equal, we trust PIR Fused Location
  • 18. 18 • Facts now represent the state of each distinct sensor – Instead of the history of sensor events – To detect state change, previous state is also stored • Changes are necessary only for rules at the lowest level – Second and higher-level rules remain unchanged • Transparent to the way sensor events are generated • Any state change event is linked to a related atomic activity – Holds for DC sensors, electricity and water flow meters – Rules fire only once when sensor values change – no need for “clean up” rules From Offline to Real-time
  • 19. 19 • Each consecutive activation/deactivation of a PIR sensor corresponds to the user being in the associated room • Subsequent activations extend the user’s stay when – Activation directly follows the last deactivation – The elapsed time between them does not exceed a threshold – No activation has taken place in a different room in the meantime • State-based approach is not applicable for video sensors – Video cameras do not broadcast a single value • Each reported bounding box is stored briefly – Combined to create facts that represent a period of stay in a room – The same heuristics used for ghost detection Online Localisation (1)
  • 20. 20 • Each time a PIR sensor is activated, the system fuses available information to decide on its validity – If there is no active video sequence and no activity detected, we assume PIR is valid – If the active video sequence with the highest confidence agrees with PIR, we conclude the user is in the room, summing confidence values – If video reports a different room, we assume the user is in the room where the most recently recognised atomic activity was performed Online Localisation (2): Fusion
  • 21. 21 • Offline and real-time versions implemented in Java, using Jess as a rule engine – Implemented rules designed to accommodate variable reporting periods • Real-time version built as an MQTT client – Sensor messages are broadcast in separate threads • 10 participants executed an ADL script of half-hour duration, twice, in the SPHERE house – Ground-truth data acquired through annotation of video images collected using a head-mounted camera – Subset of performed activities that are recognised: interaction with doors, electrical devices and water taps, preparing a snack/drink, washing hands/dishes, brushing teeth, bathing/showering Implementation and Data Collection
  • 22. 22 • TP (true positive): activity performed and recognised • FP (false positive): activity not performed but recognised • FN (false negative): activity performed but not recognised • 𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = 𝑇𝑃 𝑇𝑃+𝐹𝑃 %, 𝑟𝑒𝑐𝑎𝑙𝑙 = 𝑇𝑃 𝑇𝑃+𝐹𝑁 % Evaluation Results
  • 23. Outline 23 Introduction • Activity Recognition • The SPHERE Project • Related Work Rule-based ADL Recognition • Offline Version • Online Version • Experimental Evaluation Conclusions & Future Work
  • 24. Concluding remarks • Rule-based system capable of operating on both historical and real-time, multi-modal sensor data acquired in a smart home – Bottom-up, multi-level rules to support complex ADL scenarios – Non-deterministic patterns to account for missing activities • Sensor fusion and heuristics to achieve robustness against noise – 95% recall and 88% precision on average for a significant subset of activities – 93% room-level localisation accuracy due to effective ghost detection and location fusion rules 24
  • 25. 25 • Integrate wearable sensor data – Infer activities unidentifiable with only the other sensors – Improve localisation accuracy or provide an alternative to video cameras when they are not available/allowed • Explore multi-inhabitant scenarios – Use localisation results to pin down activities to the person performing them – For some activities, localisation needs to be more fine-grained than room-level • Explore hybrid approach with Machine Learning research within SPHERE – Incorporate rules as features in ML algorithms – Use rules that act on the results of ML algorithms – Devise ML techniques to learn rules Current and Future Work
  • 27. 27 • Chen, L., Nugent, C.D., Wang, H.: A knowledge-driven approach to activity recognition in smart homes. IEEE Trans. Knowl. Data Eng. 24(6), 961–974 (2012) • Filippaki, C., Antoniou, G., Tsamardinos, I.: Using constraint optimization for conflict resolution and detail control in activity recognition. In: Keyson, D.V., Maher, M.L., Streitz, N., Cheok, A., Augusto, J.C., Wichert, R., Englebienne, G., Aghajan, H., Krose, B.J.A. (eds.) AmI 2011. LNCS, vol. 7040, pp. 51–60. Springer, Heidelberg (2011) • Helaoui, R., Riboni, D., Stuckenschmidt, H.: A probabilistic ontological framework for the recognition of multilevel human activities. In: Mattern, F., Santini, S., Canny, J.F., Langheinrich, M., Rekimoto, J. (eds.) UbiComp 2013, pp. 345–354. ACM (2013) • Meditskos, G., Dasiopoulou, S., Kompatsiaris, I.: MetaQ: a knowledge-driven framework for context-aware activity recognition combining SPARQL and OWL 2 activity patterns. Pervasive Mob. Comput. 25, 104–124 (2016) • Skarlatidis, A., Paliouras, G., Artikis, A., Vouros, G.A.: Probabilistic event calculus for event recognition. ACM Trans. Comput. Log. 16(2), 11:1–11:37 (2015) • Woznowski, P., Fafoutis, X., Song, T., Hannuna, S., Camplani, M., Tao, L., Paiement, A., Mellios, E., Haghighi, M., Zhu, N., et al.: A multi-modal sensor infrastructure for healthcare in a residential environment. In: 2015 IEEE International Conference on Communication Workshop, pp. 271–277. IEEE (2015) References

Notes de l'éditeur

  1. SBG/SVG/SEG: Raspberry Pis SHG: Linux server Environmental sensors: Libelium nodes