The Vigilant Home
Machine intelligence doesn’t have to be humanoid.
Most robots are extremely costly, a chief reason it could be years before mobile assistive machines are deployed to help people recovering from strokes or living with other impairments. But related technologies are nearer to fruition. “Robots have to be charged frequently, and they aren’t always the most appropriate technology for the home,” says Martha E. Pollack, dean and professor at the University of Michigan’s School of Information. “If there’s stuff on the floor, robots can’t navigate very well, and a house with two stories will require two robots. But sensors embedded in the environment combined with machines’ learning to process the data have come of age and may be more feasible in the short term.”
Carnegie Mellon’s Quality of Life Technology Project is working on such gadgetry, including a device that is attached to a walker to direct people with cognitive disabilities, to monitor their walking and to encourage them to slow down if they’re tired. Another prototype recognizes and adjusts for the tremors of Parkinson’s disease to help people with the disorder steer a car.
Pollack’s research involves devices that can assess the status of a cognitively impaired person living independently. In one test of this technology, graduate students wearing bracelets that read RFID tags—those radio frequency identification bar codes that help track shipped packages—were asked to make coffee using objects such as measuring spoons and coffee cans with RFID markers. Based on the order in which the students touched the objects and how long they took with each, the technology was able to establish an “object-use fingerprint” for each student, making it possible to detect deviations in normal patterns—to alert a caregiver, for example, that a person may be getting disoriented.
People with cognitive problems already act in unpredictable ways, however, so Pollack has conducted a trial with 25 patients suffering from traumatic brain injuries to determine whether the sensors can recognize meaningful changes in irregular behavior. “This technology may give caregivers an earlier, more finely tuned indication that a person’s cognitive abilities are changing,” says Pollack, who is analyzing raw data from the study and isn’t yet able to characterize the results or draw conclusions. “And these devices are much less obtrusive than home-monitoring solutions involving cameras mounted around the house. Having caregivers get encrypted data once a day about someone’s routines could be valuable without violating anyone’s privacy.”