The Independent LifeStyle AssistantTM(I.L.S.A.): AI Lessons LearnedIn IAAI 04, San Jose CA, July 25-29, 2004, pages 852-857.Karen Zita Haigh, Liana M. Kiff,Janet Myers, Valerie Guralnik, Christopher W. Geib, John Phelps, Tom WagnerHoneywell Laboratories, 3660 Technology Drive, Minneapolis, MN 55418{karen.haigh,liana.kiff}@honeywell.comAbstractThe Independent LifeStyle AssistantTM(I.L.S.A.) is an agent-based monitoring and support system to help elderly peo-ple to live longer in their homes by reducing caregiver bur-den. I.L.S.A. is a multiagent system that incorporates a uni-fied sensing model, situation assessments, response planning,real-time responses and machine learning. This paper de-scribes the some of the lessons we learned during the devel-opment and six-month field study.1 IntroductionHistorically, 43% of Americans over the age of 65 will en-ter a nursing home for at least one year. We have beendeveloping an alternative: an automated monitoring andcaregiving system called Independent LifeStyle AssistantTM(I.L.S.A.) [9; 10; 12]. Researchers and manufacturers aredeveloping a host of home automation devices that will beavailable in the near future. I.L.S.A.’s concept is to integratethese individual devices, and augment them with reasoningcapabilities to create an intelligent, coherent, useful assistantthat helps people enjoy a prolonged, independent lifestyle.From January to July 2003, we field tested I.L.S.A. in thehomes of eleven elderly adults. The I.L.S.A. field test wasdesigned to complete an end-to-end proof-of-concept. It in-cluded continuous data collection and transmission via secu-rity sensors installed in the home, data analysis, informationsynthesis, and information delivery to I.L.S.A. clients andtheir caregivers. The test concentrated on monitoring twoof the most significant Activities of Daily Life: medicationand mobility. All ADL-based monitoring was performed byfamily caregivers.This paper describes the system we built, outlines the fieldstudy, and then describes the major lessons we learned relat-ing to AI technology:• Agents• Developing and making use of an ontology• Automated Reasoning• IntegrationHaigh et al [9] describe many additional lessons learned, in-cluding client selection, system configuration, and usability.Copyrightc° 2004, American Association for Artificial Intelli-gence (www.aaai.org). All rights reserved.2 System DescriptionThe main goal of the field test was to demonstrate the com-plete cycle of I.L.S.A. interactions: from sensors to datatransmission to reasoning to alerts and home control.We selected our initial feature set based on their impor-tance ranking, the ability to exercise the full range of techni-cal capabilities of the I.L.S.A. architecture, and the need tolearn more about a particular area. The ability to implementand appropriately support a robust test application was thefinal determining factor. The system we field tested had thefollowing significant features:• Passive Monitoring: basic mobility, occupancy, medica-tion compliance, sleeping patterns.• Cognitive Support: reminders, date/time of day.• Alerts and Notifications: auto contacting caregivers (bytelephone).• Reports: summary reports of client behavior.• Remote access to information (Internet or telephone)• Control: modes (on/off).Other capabilities and features were tested in the lab.2.1 ArchitectureBecause clients age, and technology changes, I.L.S.A. had tobe rapidly deployable, easy to configure, and easy to update.To meet these requirements, we decided to use an agent-oriented approach [10]. An agent-based architecture wouldprovide modularity, distribution, functional decoupling, anddynamic discovery of capability. It would also make ourontology explicit. We selected JADE as our environment [1].The agents in the system included device controllers,domain agents, response planners, and system manage-ment. One of each of the following agents were created foreach human client: Medication, Mobility, Modes (on/off),Reminders, ResponseCoordinator, MachineLearning. Thesketch in Figure 1 shows sample capabilities.Exactly one each of the following agents was cre-ated on the server (one server for all the human clients):PhoneAgent, Platform, Database. In the research system,we explored task tracking (Section 5.1), machine learningtechniques (Section 5.3) and several more domain agents.The hardware employed in the I.L.S.A. field test consistedof readily available Honeywell home automation and controlproducts. The Honeywell Home Controller served as thebackbone for communicating sensor events out of the home.Figure 1: The Mobility agent tracks the client’s activity.2.2 Field Study EnvironmentsWe installed I.L.S.A. in the homes of four system engineers,and eleven elderly clients. From July 2001 through Decem-ber 2001 we installed four systems in engineers’ homes, andfocused on hardware configuration to determine which sen-sors were most effective. No reasoning components or userinterfaces were included in these deployments.Beginning January 2003, we installed I.L.S.A. into thehomes of eleven elders and collected data through July 2003.We limited the number of sensors in the elders’ homes forreasons of cost and concerns about privacy—for example,it would have been difficult to find appropriate test subjectswho would accept a system with a toilet flush sensor. Eachtest home had from four to seven sensors, including onemedication caddy and several motion detectors. Two instal-lations had a contact switch and pressure mat at the exit door.2.3 User InterfaceElderly clients were equipped with Honeywell Web PadsTMwith wireless access to the Internet over a broadband con-nection. Through the Web interface, the elders could displayreminders, medication schedules & status, mobility sum-mary, on/off controls and information about their caregivers.Figure 2 shows a sample web page for the elderly client.I.L.S.A. could also deliver reminders to the elder by phone.Caregivers could access I.L.S.A. data about theirclient/family member with their normal ISP Web connec-tion. The caregiver Web interface allowed the caregiver toview and acknowledge alerts, view general ADL status (in-cluding historical trends for medication and mobility, viewand edit prescription and medication schedule, and set upFigure 2: A sample webpage from the elder user interface.scheduled reminders and personalized activity alerts.Alerts and reminders could be delivered by telephone.