These two datasets represent sensor events collected in the CASAS smart apartment testbed at Washington State University. In both sets of data, ambient sensor readings are collected while 20 participants performing five ADL activities in the apartment. This resource is valuable for designing and validating activity recognition algorithms. Further, this resource provides data for detecting errors that are helpful in assessing and intervening for functional independence. Note: Other CASAS smart home and smartwatch datasets are also available, look for more at https://zenodo.org/communities/casas. In the adl_noerror dataset, the five tasks are: Make a phone call. The participant moves to the phone in the dining room, looks p a specific number in the phone book, dials the number, and listens to the message. The recorded message provides cooking directions, which the participant summarizes on a notepad. Wash hands. The participant moves into the kitchen sink and washes his/her hands in the sink, using hand soap and drying their hands with a paper towel. Cook. The participant cooks a pot of oatmeal according to the directions given in the phone message. To cook the oatmeal the participant must measure water, pour the water into a pot and boil it, add oats, then put the oatmeal into a bowl with raisins and brown sugar. Eat. The participant takes the oatmeal and a medicine container to the dining room and eats the food. Clean. The participant takes all of the dishes to the sink and cleans them with water and dish soap in the kitchen. In the adl_error dataset, a scripted error is introduced. The errors are: Make a phone call. Error: The participant initially dials the wrong number and has to redial. Wash hands. Error: The participant does not turn the water off after washing his/her hands. Cook. Error: The participant does not turn the burner off after cooking the oatmeal. Eat. Error: The participant does not bring the medicine container with them to the dining room. Clean. Error: The participant does not use water to clean the dishes. The files are named according to the participant number and task number (e.g., p01.t1.csv contains sensor data for participant 1 performing task 1). There is one sensor reading in each row with fields date, time, sensor, and message. A floorplan of the smart apartment is provided in Chinook.png, together with the locations of the sensors. A zoomed-in look at the Chinook cabinet with sensors is provided in Chinook_Cabinet.png. The sensors are categorized (and named) as: M01 - M026: PIR motion detectors (ON when detected motion starts and OFF when it stops) I01 - I08: item use sensors for (in order) oatmeal, raisins, brown sugar, bowl, measuring spoon, medicine container, pot, phone book (PRESENT or ABSENT indicating item is on sensor or not) D01: door sensor on kitchen cabinet (OPEN or CLOSE) AD1-A and AD1-B: water sensors for kitchen sink (value indicates level) AD1-C: burner sensor (value indicates level) asterisk: phone use sensor

Citation: Please cite the following paper when using this dataset:Cook, D. & Schmitter-Edgecombe, M. (2009). Assessing the quality of activities in a smart environment. Methods of Information in Medicine, 48(5):480-485. https://doi.org/10.3414/ME0592