Page 1 ABSTRACT In recent years
In recent years, there is a drastic growth in the human population which resulted in huge cost of
living and sophisticated jobs. Employees are unable to take care of their old parents, who mostly
live in the country side. Because of this the old aged pupils are left unmonitored and often no one
to care for them.
A study revealed that most of the elder people and old aged people die because of immediate
response when they met with accidents. These accidents are of minor damage, but because of them
not being monitored, even there small accidents at home are becoming a cause for their death.
By god’s grace there is a proportional growth in the field of science and technology. Scientists and
engineers invented devices to monitor the people who are prone to these kinds of accidents. Most
of the accidents commonly occur with a fall of the person. The devices in this case are known as
fall detectors, which sense a fall of the person.
Fall detectors can be classified into different types depending on the sensors used, the algorithms
involved, the method of detecting the fall, and few other factors.
Keywords: accelerometer, Arduino, Bluetooth, algorithm
LITERATURE SURVEY REPORT
In a research paper “Automatic Detection of Human Fall in Video”, authored by Vinay Vishwakarma, IIT kharaghpur, the fall detector used was of video based. Human activity is captured and further analyzed using image processing techniques. The fall detector in this case used an adaptive background subtraction method using a Gaussian mixture model. Two fall models are used, they are: fall detection, fall conformation. It uses a two-state finite state machine for the detection and confirmation of fall.
In a research paper named “A Multi-Camera Vision System for Fall Detection
and Alarm Generation” authored by R. Cucchiara, University of Modena and Reggio Emilia introduced a multi camera vision system for fall detection, tracking people, recognizing dangerous behavior. In such situations a suitable alarm, can be sent by means of SMS. The multi camera system, objects are extracted from each single camera module by using background suppression.
In a research paper “Human Fall Detection Using Kinect Sensor” by Michal Kepski, a fall detector using kinect sensor is introduced. The device of concern takes depth reference images, and the distance between the person and the ground plane is calculated. RANSAC algorithm is used to calculate the ground plane’s depth. Hough transform, NITE based skeleton tracking is used.
In a research paper “A Simply Fall-Detection Algorithm Using Accelerometers on a Smartphone” by Ekachai Thammasat, Thailand Institute of Scientific and Technological Research a fall detector using the tri-axial accelerometer in the phone is introduced. Tri-axial accelerometer delivers the x, y; z coordinates of the device or indirectly that of the person to the microcontroller. So, when the person falls there is a spike generated in the x, y, and z axes, which then is detected by the system.
In a research paper “Fall Detection using Doppler Radar and Classifier Fusion” by Liang Liu a fall detector using dual Doppler radar system is used to generate specific Doppler signature for each type of movement. This is then categorized as a fall or a non-fall with the help of Choquet integral that combines partial decision information from each sensor and forms a final fall/non-fall decision.
A device for the elder people which they have to wear in order to get help when they fall as soon as possible. The device must be cost and space effective. It must be a low power consumption device.
The device must deliver a message to the emergency services when the fall occurs.The device must be wireless.
This paper presents a cost-effective fall detection system using Arduino, which uses an accelerometer and a Bluetooth module in order to detect a fall and alert the emergency services respectively.
The device is a low power consumption system ans can be worn as a chest belt.
In the past decade, the population in the world has been increasingly aging. Korea, for example, is rapidly changing into an “aging society.” The elderly, especially those above the age of 65, are exposed to falls owing to the deterioration of their physical functions. When an elder person falls and becomes unconscious or is unable to move his/her body, he/she may succumb to the injuries that caused the fall. Thus, research and development of a system that can automatically detect falls in the elderly or other patients has been actively studied.
Because of the expansion of the Internet in the 90s, it is now commonly referred to as the Internet of Things (IoT). The pervasive and seamless interaction among objects, sensors, and computing devices is an important concern of the IOT. Smart embedded objects such as a fall-detection sensor with wireless communication will also become an important part of the IoT.
The identified fall-detection systems can be classified primarily into two categories: context-aware systems and wearable devices. Context-aware systems use devices such as cameras, floor sensors, infrared sensors, microphones, pyroelectric infrared (PIR) sensors, and pressure sensors, deployed in the environment, to detect falls. Their principal advantage is that a person is not required to wear any special equipment. Wearable device-based approaches rely on clothing with embedded sensors to detect the motion and location of the body of the subject. The advantages of wearable devices are the cost efficiency, ease of installation, setup, and operation of the design.
There are two main approaches (algorithms) to detect falls: simple threshold and machine learning methods. In the simple threshold method, threshold values of specific parameters calculated from sensor data such as 3-axial acceleration are used to detect a fall. Automatic fall detection using a threshold-based method of single parameters, calculated using acceleration measured by an accelerometer, has a high sensitivity (about 100%); however it has a relatively low specificity. Automatic fall detection using multiple parameter combinations has a relatively high sensitivity (85.7%) and specificity (90.1%). Automatic fall detection using angular velocities measured using a gyroscope has a high sensitivity (100%) and specificity (97.5%). Further, automatic fall detection using multiple parameters that are calculated using the acceleration and angular velocities measured by an accelerometer and a gyroscope, respectively, has a high sensitivity (91%) and
specificity (92%). They are simple to implement and their computation effort is minimal. However, they have a problem with the tolerance of individual behaviour and are less accurate for detecting falls that occur. In the machine learning method, various types of fall and activity of daily living (ADL) patterns are trained by a learning algorithm and then an event is classified as a fall or ADL by applying it to an evaluation algorithm. The machine learning methods include support vector machine (SVM), Gaussian distribution of clustered knowledge, decision tree, and hidden Markov model (HMM). The machine learning method is more sophisticated and leads to better detection rates with accuracy of over 95%. Unfortunately, it is difficult to implement the machine learning approach due to the heavy computational and resource requirements. The combination of the two approaches for fall detection has not yet been investigated.
In this paper, a fall-detection algorithm using 3-axis acceleration is proposed. The fall-feature parameters, calculated from the 3-axis acceleration, are applied to a simple threshold method. Then, the falls that are determined from the simple threshold are applied to the HMM to distinguish between falls and ADLs. The results from a simple threshold, HMM, and the combination of the simple method and HMM are compared and analyzed.1
In Fig. 1 a diagram of the proposed system is shown. This consists of an electronic circuit attached to the thorax of the person, with a MEMS (Microelectromechanical Systems) acceleration sensor capable to detect falls and impacts. In addition, a processor to constantly read the acceleration data is available to determine fall events. In such cases, the processor sends AT commands via a Bluetooth link to a local mobile phone (near the elder). These commands instruct the local phone to generate SMS alerts by the public cellular network to previously selected recipients.
Figure 1:Working diagram
This comprises a three-axis MEMS accelerometer,Analog Devices ADXL345, and an Atmel ATMEGA88L microcontroller, both surface mounted and with low power consumption. The MEMS measured accelerations in the range of +/- 16g (g = 9.8m/s2). The microcontroller used is mid-range, since the proposed system does not require complex computation or enormous storage capabilities. The connection between the sensor and microcontroller is through the I2C digital serial bus. The sensor was programmed to acquire analog samples of acceleration, in each of its three axes, at a sampling frequency of 200Hz (every 5ms). All samples were converted to 13-bit digital values. The sensor includes a digital filter to reduce internal noise. Additionally, the circuit also contains noise decoupling capacitors. The programming and configuration of the microcontroller was performed in C language and included communication routines for the I2C bus and to access the ADXL345 sensor. Other routines were also implemented for the RS-232 prt communication required for sending acceleration data to an application developed in Matlab, during the system test stage.
FALL DETECTION LOGIC
The fall detection algorithm was based on the use of an acceleration threshold in order to discriminate whether an event is or is not a fall. Continuous monitoring of acceleration was performed by the sensor taking into account the different positions of the person before, during and after a fall. The state of rest or ADL (activities of daily living) was considered as the normal state of the elderly (e.g., sitting in a chair or walking) with low magnitudes of acceleration (less than 4g). When the old man loses his balance, goes to a momentary state of free fall with an acceleration similar to Earth’s gravity and then hit the ground. This impact is characterized by its duration (around 30ms) and the peak magnitude of acceleration (more than 4g). After the impact, it returns to a quiescent state with the person in a stunned or prolonged stationary position lying horizontally on the ground. The detection of this state also confirms the fall event.
To ensure a system with a high reliability, a medical grade, Class 1, 100m range, highly immune to interferences, an Ezurio BTM402 Bluetooth module was included. This device was used for sending AT commands from the processor to the local phone during the fall events. The local phone receives commands and automatically generates text messages to the programmed recipient. SMS is used because of its high reliability and fast delivery of messages, even with overloaded telephone networks.
Figure 2: Posture vs axes
ADI’s digital triaxial accelerometer ADXL345 is the motion sensor used in this system. Arduino is the main brain of this system. The HC-06 Bluetooth device connects the whole system to a phone nearby. The bluetooth device is a cheap and highly reliable solution but can work only for limited ranges
Each hardware component of the wearable device is working under low voltage and the detection algorithm does not need complex calculation resource, so the power consumption of the whole device is quite low. A 1200 mAh, 3.7 V polymer lithium battery is quite enough to provide the need of the wearable device for a couple of days.
Figure 3: Circuit Diagram
If the person sits suddenly, the alarm may get triggered. This is a serious issue as it creates unnecessary tension among the family members and the emergency services may arrive and get to know about the situation and may lose the hope on the fall detection system.
Sometimes the elderly may not fall but may roll down the steps and may get injured seriously. This does not create the necessary spike that the accelerometer needs to detect a fall, and may miss the fall. The elderly remain un attended and may soon meet their fate!.
This is a serious concern, which must be resolved in order to make this project more robust.
This paper developed a fall detection system based on a single triaxial accelerometer based wearable device. There is no special requirement of the device’s mounting orientation because the algorithm does not claim the axes of accelerometer to be fixed strictly. The system has low power consumed hardware design and highly efficient algorithm which could extend the service time of the wearable device. Both the hardware and software designs are suitable for wearable and outdoor application.
As normal activity of resting also has similar rotation as falling, it may trigger fall alarm when the body hits ground heavily. So the choice of a threshold is quite important to distinguish falling from heavily lying activity. Sufficient sample number collected from subjects with different age and gender will improve the reliability and robustness of the threshold. Beside these, technologies such as SVM and neural network are considerable to seek out a proper classification method based on the features used in this system.
1. Karlsson M. K., Magnusson H., von Schewelov T., Rosengren B. E. Prevention Of Falls In The Elderly—a review. Osteoporosis International. 2013.
2. Shany T., Redmond S. J., Narayanan M. R., Lovell N. H. Sensors-Based Wearable Systems For Monitoring Of Human Movement And Falls. 3. Vishwakarma V., Mandal C., Sural S. (2007) Automatic Detection of Human Fall in Video. In: Ghosh A., De R.K., Pal S.K. (eds) Pattern Recognition and Machine Intelligence. PReMI 2007. Lecture Notes in Computer Science, vol 4815. Springer, Berlin, Heidelberg
4. Cucchiara, R. , Prati, A. and Vezzani, R. (2007), A multi?camera vision system for fall detection and alarm generation. Expert Systems, 24: 334-345.
5. Thammasat, Ekachai ; Chaicharn, Jarree. (2012). A simply fall-detection algorithm using accelerometers on a smartphone. 1-4. 10.1109/BMEiCon.2012.6465471.