The mobile sensing platform: An embedded activity recognition system

被引：289

作者：

Choudhury, Tanzeem ^{[1
]}

Consolvo, Sunny

Harrison, Beverly ^{[4
]}

LaMarca, Anthony ^{[4
]}

LeGrand, Louis ^{[4
]}

Rahimi, Ali

Rea, Adam

Borriello, Gaetano ^{[2
]}

Hemingway, Bruce ^{[5
,6
]}

Klasnja, Predrag Pedja ^{[7
]}

Koscher, Karl ^{[7
]}

Landay, James A. ^{[8
]}

Lester, Jonathan ^{[7
]}

Wyatt, Danny

Haehnel, Dirk ^{[3
]}

Hightower, Jeffrey

机构：

[1] Dartmouth Coll, Hanover, NH 03755 USA

[2] Univ Washington, Seattle, WA 98195 USA

[3] Stanford Univ, Dept Comp Sci, Stanford, CA 94305 USA

[4] Intel Res Seattle, Seattle, WA USA

[5] Univ Washington, Dept Comp Sci, Seattle, WA 98195 USA

[6] Univ Washington, Baxter Comp Engn Lab, Seattle, WA 98195 USA

[7] Univ Washington, Seattle, WA 98195 USA

[8] Univ Washington, Dept Comp Sci & Engn, Seattle, WA 98195 USA

来源：

IEEE PERVASIVE COMPUTING | 2008年 / 7卷 / 02期

关键词：

Activity recognition; Embedded systems; Machine learning; Wearable computers;

D O I：

10.1109/MPRV.2008.39

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

The Mobile Sensing Platform (MSP) is a small-form-factor wearable device designed for embedded activity recognition. The MSP aims to support context-aware ubiquitous computing applications. Activity recognition systems have three main components including a low-level sensing module that gathers relevant information about activities using microphones, accelerometers, and light sensors, a feature processing and selection that processes the raw sensor data into features that help to discriminate between activities, and a classification module that uses the features to infer the activity of particular individual. Activity recognition requires sensor data to be processed and classified into meaningful activity chunks and involves extraction of various feature that serve as an input to the classification module. The MSP runs standard Linux with complete support for multitasking, IP networking via Bluetooth, or USB, and file flash system.

引用

页码：32 / 41

页数：10

共 15 条

[1] Activity recognition from user-annotated acceleration data [J].

Bao, L ;

Intille, SS .

PERVASIVE COMPUTING, PROCEEDINGS, 2004, 3001 :1-17

[2]

Consolvo S., 2008, P ACM SIGCHI C HUM F

[3]

Kern N, 2003, LECT NOTES COMPUT SC, V2875, P220

[4] Unsupervised, dynamic identification of physiological and activity context in wearable computing [J].

Krause, A ;

Siewiorek, DP ;

Smailagic, A ;

Farringdon, J .

SEVENTH IEEE INTERNATIONAL SYMPOSIUM ON WEARABLE COMPUTERS, PROCEEDINGS, 2003, :88-97

[5]

Lester J, 2006, LECT NOTES COMPUT SC, V3968, P1

[6]

LIAO L, 2007, P INT JOINT C ART IN

[7] Extracting places and activities from GPS traces using hierarchical conditional random fields [J].

Liao, Lin ;

Fox, Dieter ;

Kautz, Henry .

INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH, 2007, 26 (01) :119-134

[8]

LUKOWICZ P, 2002, LNCS, V2498, P361

[9]

MAHDAVIANI M, 2007, P NEUR INF PROC SYST

[10]

Maurer U, 2006, LECT NOTES COMPUT SC, V3864, P86

← 1 2 →