- Movement pattern detection can be applied in a variety of applications such as assisting independent living of seniors at home, behaviour understanding in surveillance systems, sports analytics, and robotics. This project develops a scheme that fuses information from diﬀerent sensors to detect movement patterns. This report contains three main parts: information collection and processing, pattern detection using the information collected, and algorithm implementation and results. The information needed for movement pattern detection comes from an inertial measurement unit (IMU) and a barometer. The information from the accelerometer and the gyroscope is ﬁrst combined by using a complementary ﬁlter. The measurements in the body coordinates of the IMU are then transformed into data in the earth coordinates via quaternions. We then develop a scheme that exploits the advantages of the vupport vector machine and the k-nearest neighbor algorithm for motion detection. These schemes are ﬁnally implemented to detect four diﬀerent movement patterns: walking, running, standing up and falling down, which are classiﬁed into static and dynamic motions. For dynamic motion, the diﬀerence of tilt angle and height could be used to distinguish the standing-up and falling-down patterns; for static motion, the diﬀerence of velocity in the horizontal plane could be used to distinguish the walking and running patterns.