DEVELOPMENT OF WIRELESS MEASUREMENT SYSTEM USING INERTIAL MEASUREMENT UNIT AND FORCE SENSING RESISTOR FOR GAIT PARAMETER EXTRACTION

• Main Author: Adib Syamlan, Muhammad
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67550
• Institution: Institut Teknologi Bandung

Gait is one of the cyclic movements occured in the body which used to perform daily life activities. Every person has its different spatiotemporal, kinetic, and kinematic parameters that makes every person unique. Until now, many studies have already been done regarding data analysis related to the patterns of body movements. There are two methods regarding the data acquisition, a non-wearable-based and a wearable-based. Non-wearable-based require a large instrument (cameras/motion capture, force plates), meanwhile wearable-based measurements typically use more than one sensor (small, lightweight) placed on the surface of human body. Some advantages of wearable-based data acquisition are cheaper and can be used while carrying out daily life activities. Many wearable-based systems have been done before. Some existing studies use less than 7 sensors with a limit only used to analyze the pattern of lower limb movements. In addition, several studies use a nonrechargeable battery rather than a rechargeable one, which both act as a supply. In terms of data communication, there are two methods available, that is communication via cable or wireless. One of the problems regarding to wireless communication with more than one sensor module is that each data collected is not sync with each other. This unsynchronized data will only cause much more problems in the later processing stage because it is a data problem. The current existing algorithm is able to sync each of the sensor data with good accuracy, but the high computational load needs to be considered, especially in the field of Wireless Sensor Network (WSN). To overcome the problems, we design a wearable-based measurement system with 11 sensor modules placed on Right Upper Arm (RUA), Right Lower Arm (RLA), Left Upper Arm (LUA), Left Lower Arm (LLA), Body (B), Right Thigh (RT), Right Shank (RS), Left Thigh (LT), Left Shank (LS), Right Foot (RF), and Left Foot (LF). Each sensor module consists of one ESP32 microcontroller used to carry out the program routines, and also there is an Inertial Measurement Unit (IMU) sensor used to detect the body segments position based on the acceleration of the body detected by accelerometer sensor and the angular velocity detected by gyroscope sensor. The inertial sensor values are then pre-processed using LPF with a cut-off frequency (fc) of 10Hz for noise reduction, then fused by using complementary filter to get one accurate angle. In this work, 6 plantar sensors (FSR) are placed on each iv side of the shoe’s insole, which the data are collected by two sensor modules (RF and LF). FSR used to detect the Ground Reaction Force (GRF) of the subject. GRF data are useful to detect spatiotemporal parameters of gait. In this work, we use 3.7V LiPo battery to act as a supply for the whole circuit. This battery is connected to HT7833 voltage regulator before being connected to pin 3.3 ESP32. Based on the results, ESP32 is able to work well for ~2 hours 25 minutes. This duration is considered good compared to the use of 9V battery, which ESP32 works properly only for ~11 minutes. Other advantages of LiPo battery besides its long-term usage, it has a small dimension, can be recharged (not a one-time use). This design also low cost (Rp. 3.393.500,00) and can be used for long term experiment. In this work, each sensor module sent the sensor data using router (WiFi) to PC. Data synchronization is carried out using a star topology (master-slave) with nonrealtime data transmission. In this method, the PC acts as a master which function to send time parameter to each of the sensor modules which acts as a slave. In the end of the experiment, slave will send back time parameter for data synchronization process to PC. By using this method, we are able to minimize the delay in the order of tens/hundreds of ms, so this method is still less accurate when compared to existing studies. However, this method has less computational load because the synchronization method only requires one variable data storage in the ESP32 memory, so it is suitable to synchronize non-realtime data with microcontroller with a low memory capacity. The results of the sync data are used as an input for extracting gait parameters. By using two FSR sensors placed on one side of the shoe’s insole, we are able to extract 21 total parameters consisting of 20 parameters extracted by sensors placed on one side of the foot (right or left) and 1 parameter is calculated based on sensor data from both sides of the foot. When compared with existing studies, we are able to get a lot of parameters by only using 2 FSR sensors.