Developing Real-time Automatic Step Detection On A Low-Cost, Performance-Constrained Microcontroller.

The rapid urbanization process in recent times has led to the rapid deterioration of the structures of buildings, posing a risk of explosion, collapse, and rescue challenges for firefighters. To minimize risk, a system capable of locating search and rescue personnel is essential. This research proposes the use of inertial sensors and decision trees algorithm to estimate the number of steps taken by a moving individual. Step-counting algorithm is a popular method to determine the position of a moving object in 3D, but its implementation on microcontrollers faces performance and memory challenges. A solution to this problem is to use decision tree to increase performance and reduce memory requirements, making it compatible with real-time data and low-power consumption microcontrollers. This system is not dependent on pre-setting conditions. The test results from public and private data sets show that the proposed method can detect steps with a high accuracy of 99%.