Reliable Real-Time Recognition of motion related human activities using MEMS inertial sensors

Abstract

Knowledge about the current motion related activity of a person is information that is required or useful for a number of applications. Technical advances in the past years have reduced prices for sensors capable of providing the necessary input, in particular MEMS based inertial measurement units (IMUs). In addition to a low price, unobtrusiveness is a requirement for a activity recognition system. We achieve this by mounting one IMU to the belt of the user. In this work we present the design of our recognition system, including the features computed from the raw accelerations and turn rates as well as four different classification algorithms. These are used in Bayesian techniques trained from a semi naturalistic, labeled data set. The best classifier recognizes the activities ’Sitting’, ’Standing’, ’Walking’, ’Running’, ’Jumping’, ’Falling’ and ’Lying’ of any person with recognition recalls and precisions between 93 and 100% except for an only 80% recall rate for ’Falling’ as that suffers from its very short duration.