Download presentation
Presentation is loading. Please wait.
Published byBenjamin Walton Modified over 9 years ago
1
Smart Environments for Occupancy Sensing and Services Paper by Pirttikangas, Tobe, and Thepvilojanapong Presented by Alan Kelly December 7, 2011
2
2 Smart Environments and Location ► Smart spaces provide location-based services Challenges ► Assigning place names and a naming ontology ► Identifying observed people and objects ► Computing accurate location of observations
3
3 Location and Smart Environments ► Sensors observe physical phenomena Challenges ► Fusing observations from multiple sensors ► Removing noise and interference ► Compensating for environmental variations
4
4 Infrared Location Detection ► Each person wears transmitter badge ► Fixed receivers report to central server ► Limitations Very short range Line-of-sight needed Fluorescent lighting and direct sunlight interfere
5
5 Ultrasound Location Detection ► Active Bat Bat (transmitter) on person/object sends pulse Fixed receivers report to central server Uses time-of-flight trilateration ► Cricket Object is the receiver and does the calculations Uses TDOA between ultrasound and RF ► DOLPHIN - distributed positioning algorithm
6
6 RF Location Detection ► Frequency Modulation (FM) Signal strength between FM radio stations ► Wi-Fi Signal strength between access points Accuracy depends on AP density and mapping ► Ultra-wideband (UWB) Very precise measurement of UWB radio pulses Lower sensor density necessary
7
7 Vision Location Detection ► Cameras track persons or objects Motion Body parts (by color) Face detection or recognition
8
8 Pressure Location Detection
9
9 Location Estimation Algorithms ► Occupancy sensing provides abstract information about a user’s place Movement, and/or Static position, and/or Relative distance to other objects ► Bayes filtering Noise indicates most probable state Algorithm estimates angle and distance
10
10 Bayes Filtering Algorithms ► Kalman filter Used for tracking moving objects 3 extended Kalman models ► Position ► Position-Velocity ► Position-Velocity-Acceleration
11
11 Bayes Filtering Algorithms ► Particle Filter Estimates location at given time Builds a particle cloud — a distribution cloud of a finite number of (position, probability) pairs
12
12 Routine Learning ► Days/weeks/months of observations ► Identification of critical places ► Naming or geo-coding of these places ► From data, algorithm can predict path ► Then, smart services can be provided Location-based reminders Advice based on next step of learned routine
13
13 Platform: EasyLiving ► Microsoft Research ► Tracks person and their interaction with system Computer session can follow user to a new device Local lights, speakers, etc. turned on and off ► Sensors 3D stereo cameras Pressure mats Thumbprint reader Keyboard login
14
14 Platform: Aware Home ► Georgia Institute of Technology ► Research focused on evaluating user experiences in the home domain ► Sensors Ceiling cameras RFID floor mat system Door lock fingerprint readers Voice recognition
15
15 Conclusions ► Authors: there is no one ‘perfect’ occupancy sensing system Accuracy Privacy User preference Cost ► Authors: Next steps are to accurately predict users’ actions ahead of time
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.