Drop Without DangerOctober 1, 2006 By: Barbara G. Goode, Sensors Sensors
Engineer of the Year 2006 winners make a clever sensor application to solve a vexing portable electronics problem.
Inspired by the tragic loss of a young engineer whose life and potential were cut short (sidebar, Honoring Eamonn), Sensors last year announced the launch of its Engineer of the Year award. After reading and weighing the relative merits of all submissions, the editors of Sensors decided on a winner—or rather, a group of winners. Let us tell you about the inspiration for this team's work, as well as the system and method they devised to answer the challenge.
Seeking a Solution
Portable electronic devices, including computers, PDAs, cellular phones, and music players, are often accidentally dropped, and when that happens the resulting impact can damage sensitive components. Sometimes the devices can better withstand an impact if put into a protective mode; for those that include hard drives, for instance, the drives can be parked to reduce such results as data loss or disk damage. In these devices, it is beneficial to detect when a fall is actually occurring—but fall-detection methods have heretofore had limited success. Some methods have relied upon complex angular velocity calculations, for example, and are applicable only to devices with spinning disks. Other methods have been limited in their ability to detect falls accompanied by other motions such as rotation or additional external force (e.g., a nonlinear fall).
In an effort to solve the problem of free-fall detection that laptop computer manufacturers and others had struggled with, Freescale Semiconductor Inc.'s Inertial Sensor Applications Team—consisting of Michelle Kelsey, Leticia Gomez, Rod Borras, and Akihiro Ueda—got directly involved and went to work using lower-g (1.5 – 6 g) accelerometer products to determine how accelerometers can detect free fall, and by developing and testing free-fall software algorithms.
Evolution of the System
The Inertial Sensor Applications Team initially found an algorithm able to detect linear free fall without the need for an initial system calibration. A key physical property involved in such an algorithm is that under static conditions and normal motion, one or more of the three axes will always be affected by gravity; during a linear free-fall event, all three axes return to a "zero g" condition. By summing the squares of the three axial acceleration readings, it is possible to determine whether an object is static or moving.
Another step in the development of the algorithm was finding a way to detect rotational free-fall events, wherein many objects can fall, or be thrown, with a rotational component to the motion. Furthermore, by performing an analysis of the conditions immediately preceding a free-fall condition, it can indicate whether the corresponding object was dropped or thrown. In this way, the algorithm can reveal—for the purpose of product warranty claims, for instance—from what height an object was dropped, and whether it was mishandled.
The team has developed an improved system and method (patents pending) that provides fall detection in a reliable and efficient manner even in the presence of other motion. The system includes multiple accelerometers and a processor: The accelerometers provide measurements describing acceleration in all directions to the processor. The processor receives the data and compares them to a value range to determine whether the device is currently falling. If the acceleration measurement combinations are within the value range and are smooth, then a nonlinear fall is occurring. The processor then sends a signal to the device, which can take action to reduce the potential negative results of the impending impact. For example, the device can suspend operation and/or save data, thus preventing data loss if it happens to be in write mode.
Most Read Articles