UC BERKELEY UC DAVIS UC MERCED UC SANTA CRUZ Summary and New Research Directions Ruzena Bajcsy, director of CITRIS http://www.citris-uc.org 1 Outline CITRIS Overview New Research Directions 2 Initial Applications (1) Energy Efficiency Disaster Response and Homeland Defense Education 3 Initial Applications (2) Monitoring Health Care Environmental Monitoring Transportation Planning 4 February 2003 February 2001 February 2000 August 2001 February 2002 5 Mobile Sensors 6 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings 7 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure 8 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure (2) 9 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure (2) Fire and Disaster Response 10 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure (2) Fire and Disaster Response Habitat Monitoring on Great Duck Island 11 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure (2) Fire and Disaster Response Habitat Monitoring on Great Duck Island Monterey Bay Monitoring Groundwater Monitoring Taking Sensors to Market 12 Sensor Nets – A few sample projects Earthquake Response of Retrofitted Buildings Predicting Ground Failure (2) Fire and Disaster Response Habitat Monitoring on Great Duck Island Monterey Bay Monitoring Groundwater Monitoring Taking Sensors to Market Teleactors 13 SOCIETAL-SCALE INFORMATION SYSTEMS—SIS Scalable, Reliable, Secure Services Building & Using Sensor Nets MEMS for Sensor Nets CITRIS University of California 14 PicoRadio – low power sensor nodes Jan Rabaey Original technology goals » Limit power to 100 microwatt for energy scavenging » Enable energy scavenging from building vibrations Light Level Low Indoor Light Fluorescent Indoor Light Partly Cloudy Outdoor Light Bright Indoor Lamp High Light Conditions Vibration Level 2.2m/s2 5.7m/s2 Duty Cycle 0.36% 0.53% 5.6% 11% 100% Duty Cycle 1.6% 2.6% 15 PicoRadio –future goals Technology: smaller (mm3) and lower power (< 10 mwatt) Applications: “Smart surfaces”: plane wings, smart construction materials, intelligent walls How to get there? Go non-traditional! » Use non-tuned mostly passive radio’s – center carrier frequency randomly distributed » Use statistical distribution for reliable data propagation 200 mwatt Receiver 50 mwatt RF Oscillator Both in Fab 16 Energy Scavenging Paul Wright, Jan Rabaey, David Culler, … Goals: » Technology: build sensors motes that can run from solar and vibrational (piezoelectric) sources » Applications: Meet California Energy Commission goal of “energy aware houses” that can automatically measure and control energy use, based on real-time pricing Feb 10 2004 demo to CEC of temperature sensors powered by vibrations on wooden steps Jan 30 2004 demo to Assemblymember Loni Hancock 17 Common Sources of Vibrations Frequency of Peak (Hz) Peak Acceleration (m/s2) Kitchen Blender Casing 121 6.4 Clothes Dryer 121 3.5 Door Frame (just after door closes) 125 3 Small Microwave Oven 121 2.25 HVAC Vents in Office Building 60 0.2-1.5 Wooden Deck with People Walking 385 1.3 Bread Maker 121 1.03 External Windows (size 2ftx3ft) next to a Busy Street 100 0.7 Notebook Computer while CD is Being Read 75 0.6 Washing Machine 109 0.5 Second Story of Wood Frame Office Building 100 0.2 Refrigerator 240 0.1 Vibration Source 18 Tiny Temp Storage Capacitor Power Circuit Mote •3 people ran on the stairs for 40 minutes, charging capacitor •3.28 V for 816 ms • 2 temperature readings transmitted Piezoelectric Power Generator Thermistor 19 Redwoods go Wireless Todd Dawson Goal: understand role of trees in the hydrological cycle » Forests recycle ~66% of all fresh water » What is relative important of tree biology and climate in water loss? Local Context » Understand interaction of maritime fog and coastal redwoods » Maritime fog major source of water; some communities in Chile, W. Africa depend on it 20 Redwood Data Humidity vs. Time 101 104 109 110 111 Rel Humidity (%) 95 85 75 65 55 45 35 Temperature vs. Time Temperature (C) 33 28 23 18 13 8 7/7/03 7/7/03 7/7/03 7/7/03 8/7/03 8/7/03 8/7/03 8/7/03 8/7/03 8/7/03 9/7/03 9/7/03 9/7/03 9:40 13:41 17:43 21:45 1:47 5:49 9:51 13:53 17:55 21:57 1:59 6:01 10:03 Date 21 Optimizing Building Operations Ed Arens Goal: use sensor nets. to » Save energy ( up to 1/3 of US energy wasted in misheated/cooled/lit buildings) » Increase occupant comfort and productivity Obstacle: » >90% of cost of conventional sensors is wiring for power and communication Approach: » Make wireless, battery-less sensors cheap and reliable CEC goal » $20/home to put sensors in every California household • Thermostat, appliances, networked to utility for price info » Enable “demand-response pricing” 22 Seismic and Wildfire Monitoring Steve Glaser Goals » Create a “downhole” seismic array to monitor ground motion » Monitoring of wildfire to guide firefighters 23 Wild Fire Monitoring • Drop sensors with weather boards, GPS and radio •Helicopter based real-time topographic and ground cover mapping •Measured data coupled to real-time burn model •Visualization of actual and modeled observations returned to firefighters via GIS •Status: •HW, SW partially complete •grad students certified for “controlled burns” 24 Golden Gate Bridge Monitoring Greg Fenves, David Culler, Jim Demmel Monitor structural “health” » Vibrations, unusual behavior, from wind, earthquake, or local damage Challenge: high accuracy, low jitter, lots of data Two Accelerometers on prototype board: » $10, 200 mG, ADXL and $150, 30 mG, Silicon Designs 25 At the Golden Gate Bridge 26 New Research Directions Eric Brewer » Information Technology for the 3rd World Alice Agogino » Lighting and Medical Personalization Pravin Varaiya » Traffic Monitoring AnnaLee Saxenian » RFID and Privacy George Akerlof et al » Experimental Social Sciences Lab (XLAB) Paul Wright, Kathy Yelick et al » On-line education 27 Information Technology For the 3rd World Eric Brewer et al Who lives in the “3rd World”? » 3 - 4 B people with incomes < $2K / year » Little education » Local and disorganized markets But they spend on technology » 85% have TV, 21% telephone (Dharavi, India) How can IT help these people profitably? 28 3rd World Success Stories Eradication of Dengue Fever in West Africa » Sensors, models to time spraying of larvicide » Protects 30M people, freed 100K sq miles of farm land Sri Lankan Rice Market » Wireless devices to tell farmers real price of rice » Eliminated cheating by middlemen, doubled incomes Grameen Bank in Bangladesh » Microlender (to women) for businesses » Has loaned $4B, 99% recovery rate » Funded national cell phone network, profitably Can we systematically help create more of these? » Health, education, Government, commerce – many examples 29 Research Challenges Low cost, low power devices » Must be cheap, not depend on wired power Rural network coverage » 802.xx variations Low-power intermittent networking » Just when daily bus goes through village Literacy issues and User Interfaces » Assume users illiterate Shared devices and infrastructure » Data center for persistent storage 30 Partners NSF Intel, HP, HP Labs India Grameen Bank, UNDP, Markle IIT Delhi and Kanpur UC Social Scientists First deployment in India 2005 Looking for second deployment 31 Traffic Monitoring (1) Pravin Varaiya Goal » Monitor traffic to better control street lights, onramps lights, identify accidents… » Replace current technology with sensors that are • Cheaper to install (don’t dig up road for loop detectors) • Cheap to maintain • Reliable, low delay (1/10 sec at intersections) 32 Sample Traffic Data Midlane Microphone MidlaneMagnetometer Wind disturbance Current work: •Comparing network protocols •Better signal processing, especially microphone data •Collecting data on Hearst near Euclid, and Richmond FS 33 RFID and Privacy AnnaLee Saxenian, Ross Stapleton-Gray The “Sorting Door” project » Equip “doors” with RFID readers » Network them to Identification Engine (IE) » IE uses Object Name Service (ONS), other databases, to identify tags » Distribute doors in experimental environment (eg campus) Research questions » How effective are RFID readers? » How aware are people of RFID? • Ex: “Door” could announce what people are wearing or reading, record reaction » What behavioral patterns can be extracted? » What behavioral deductions can be made using IE? » Is there a “tipping point”, where RFID dramatically effects privacy? 34 What is Tele-Immersion? Tele-immersion is an emerging technology that enables cooperative interaction between geographically distributed sites. This is achieved through realistic reconstruction of the scenes in virtual space in real-time. 35 What is Tele-Immersion? Tele-immersion connects remote places through a shared virtual environment. Internet 2 3D scene Interactive 3D display at remote location UPENN Internet 2 3D scene Interactive 3D display at remote location Internet 2 3D scene Interactive 3D display at remote location UC Berkeley UIUC Bidirectional 3D data transmission 36 Applications Tele-Immersion can be used in surveillance coordinated global decision making through integrated view of the scene remote learning and training coordination of activities between distant places (physical therapy, exercise, distributed ballet, … etc) unobtrusive monitoring of people group behavior study 37 Overview of Tele-Immersion Image processing Synchronized data transmission Reconstruction These should be done in real-time with minimal delay both in computation and data transmission. 38 Tele-immersion Setup The system consists of camera clusters and display system. A camera cluster of 3 BW and 1 color cameras Teleimmersion Lab 39 How Tele-Immersion works? Image Processing Real-time depth computation fast stereo algorithm Multiple cameras to fully acquire the scene >12 camera clusters to cover 360 degrees Challenges in computer vision Texture-less regions occlusion highlights computation 3D model 3D model acquisition from multiple images 40 How Teleimmersion works? Data transmission Gigabit internet connection between multiple sites currently implemented on wired network Issues in communication bandwidth requirements (>500Mbps between two sites) (320x240x5)(byte/data) x7(stream)x10(fps) x 2 (sites) real-time constraint minimal latency and delay data synchronization • synchronization of multimedia data (video + sound) • synchronization between multiple sites 41 How Teleimmersion works? Reconstruction Reconstruction of dynamic activities in virtual space 42 Teleimmersion and Social Science Understanding group function is critically important to society and organizations. ex) group formation, diffusion of information, consensus building Technology have changed the way in which groups and teams operate. mediated meeting Face to face distance through telephone, teleconferencing, email and internet messaging 43 Teleimmersion and Social Science Mediated group research has been relatively less explored. • limited to comparison study • no systematic / integrated research established due to lack of infrastructure! • technology advances >> mediated group research Tele-immersion can bridge this gap by providing collaborating environment and infrastructure. 44 Teleimmersion and Social Science Tele-immersion can serve as • a new testbed for studying mediated group interactions • a tool for analyzing systematically human behavior • a system for constructing a large collection of database of group interactions. 45 Group meeting in Tele-immersion People meet in a shared virtual space Display Display Data Acquisition UI Data Acquisition Integrate UCB Display Display UPENN Display Data Acquisition Display 46 Group interactions in Tele-immersion Meeting in the virtual space site 1 site 3 site 2 analysis based on verbal communication (text, speech) non-verbal and interactions with technology 47 Systematic analysis of human behavior using 4D Video Stream Analysis of dynamics of human behavior • quantitative measurements of human motion and facial expression • automatic retrieval of elementary behaviors estimation of body parts from 3d point clouds 48 Database for group study Tele-immersion can construct database of a large collection of group interaction samples. • 4D dataset of audio-visual data • online repository • data resources for other analysis 49 Summary Tele-immersion is the state of art technology that combines computer vision, graphics and network communications. Tele-immersion can provide the infrastructure for studying mediated groups by establishing collaborative environment. Systematic analysis of human behavior is possible in the new environment. Tele-immersion can build database of group interactions that enables further extensive studies. 50