Author: Majumder, Apratim; Banerjee, Niladri; Nayak, Shikha; Chakraborty, B
Date published: March 1, 2011
Journal code: SNTD
A wheelchair is a chair with wheels, designed to be a replacement for walking, which can be propelled manually and also driven by motors. Wheelchairs are used by people for whom walking is difficult or impossible due to illness (physiological or physical), injury, or disability. Medically a loss of mobility can be broadly attributed to two reasons- Musculoskeletal disorders such as paralysis, paresis, ataxia, fibromyalgia, etc. and neuropsychiatrie disorders such as posttraumatic stress disorder, cognitive dysfunction etc. Motorized wheelchairs have gone through enormous modifications and technological enhancements over the last few decades. Over the years scientists have developed [1-13] various kinds of motorized-wheelchairs, some are verbally controlled by the user while some are user's mindcontrolled. The most widely available ones are computer-driven. Most of these motorized-wheelchairs available in market need human intervention to guide the chair. Either the user uses steering/handle to maneuver the chair or uses digital/computer controls  to do the same. But for the people who are very old and hence unable to move their limbs systematically to guide the chair along the desired direction or people who due to illness would fail to do so, these motorized wheelchairs  do not suffice their requirements efficiently. Hence, our endeavor in this paper was to make such a system that will overcome this impediment effectively and can help the old and ailing or the injured to find their way from their present location to a given destination within a given closed system with minimal human interference. In the case, once the user sets the chair's present location and the desired destination onto the system, the chair with the help of a unique sensor-aided automatic path-finding algorithm finds its way through all turns and avoids collision to eventually reach the destination successfully. Unlike the other intelligent motorized-wheelchairs [3-5], this system uses an effective but simplistic design and simple mechanical and electronic components and hence is largely affordable to all people, thus providing mobility-aid to all.
2.Method of Approach
The device that has been designed is a Sensor Aided Automatic Path Finding Wheelchair. It is a motorized wheelchair designed to carry one occupant at a time. It moves on its own and requires no assistance for steering from the occupant of the wheelchair or from any outside individual. It just requires information regarding the starting and stopping points of its course. Using the supplied information it can decide upon the path to take in order to complete its journey. It can detect and avoid obstructions in its path.
Since this is a motorized wheelchair whose locomotion is automatic and the course that it takes is decided by the programming that governs the movement of the chair, the entire device can be described as consisting of the sections shown by Fig. 1.
The system has been designed properly such that it can run and control the total system smoothly. The wheelchair with its facility setup and the other important part of the system, programming, algorithm and system integration has been taken up individually and designed properly. The basic part of the wheelchair consists of following parts as shown in Fig. 2:
a) The Seat This is the seat of the wheelchair where the occupant sits.
b) The Backrest. This is the backrest of the wheelchair
c) Primary Wheels: A pair of large wheels on either side of the chair that are primarily responsible for movement
d) Secondary Wheels: A pair of secondary wheels that are responsible for changing the orientation and directing the movement of the chair
e) Sensor Panels: Panels fitted with IR and sonic sensors that help in the movement of the wheelchair
* Panel 1: This is the top most panel located above the head of the passenger and consists of a set of sensors used to detect and avoid collision
* Panel 2: This is the middle most panel located at the waist region of the passenger and consists of a set of sensors used to detect and avoid collision
* Panel 3: This is the 3rd panel located below Panel 2 at the footboard of the chair and consists of a set of sensors used to detect and avoid collision
* Panel 4: This is the bottom most panel consisting of the sensors that help the chair to deduce its position in the facility while moving
* Panels 5 and 6: Located on either side of the wheelchair and used in aiding its locomotion
g) Main Circuit and Motor Box: This box houses the motors, their concerned circuits, the microcontrollers (PIC18F252) and mother board, the memory devices and other electronic circuits used to assist the movement of the wheelchair. The Microcontroller PIC18F252 is ideal for our purpose because of its enhanced flash which will be useful in processing the inputs from sensors and other peripherals. It also has 23 I/O lines to further assist the entire process. It has 1536 bytes of RAM which is again suitable for this system. There are several other microcontrollers of similar specifications are available of different makes which can also be applied.
h) User Interface: A screen and a keypad used to put in course related information into the wheelchair. Since the movements of the wheelchair are automated, it requires a guidance system to govern its movement to its destination from its starting point with efficiency and accuracy. In order to achieve this inside a closed facility, a guidance system has been constructed since common mapping technology like the GPS cannot function efficiently within a concise region. The entire facility has been broken up into a number of small points named "Bases" as shown in Figure 3. The bases are the places at which the wheelchair can stop or commence its locomotion. The wheelchair cannot stop (except upon encountering an obstacle) in the region between two successive bases under normal circumstances. It can move from one base to the next, i.e. in other words its movement has been quantized between the different bases. The bases are designated by IR emitters embedded on the floor of the facility and that once switched on emit a steady beam of IR upwards from them up towards the ceiling. Each base emits an IR beam with a different wavelength and no two are the same. Now the facility setups are integrated to make the system perform successfully when a passenger is seated on the wheelchair. In the user interface the two points, the present position of the wheelchair and the destination of the wheelchair are keyed in using the keypad and screen. The present position of the wheelchair will always be the Base number of the base that it has last crossed. The destination of the wheelchair will always be the Base number of the base that it has to reach. Once the data has been punched in, the wheelchair commences its movement forward.
For a safe journey, the wheelchair can accelerate at 0.14 m/s2 and can achieve a constant speed of 0.73152 m/s (i.e. 2.4 ft/s) in 5s. This is similar to the statistics of a wheelchair pushed by hand. As it moves forward, it moves over the IR emitters of each base. Only when it detects the IR emitter corresponding to its destination, it decelerates to a halt.
The base point IR emitters continuously emit an IR beam in wall-to-wall direction, i.e. cutting through the line of movement of the chair. The Base Detection circuit of the wheelchair consists of the blocks shown in Fig. 3.
Let the wheelchair crosses the Base with Base number 6. As soon as either (or both) of the Sensor Panel 5 or 6 located at the sides of the chair crosses directly over the IR emitter of the base, the IR beams strikes the photodiode. The photodiode produces a voltage signal by photovoltaic effect particular to the wavelength of the emitter. This signal is processed through an ADC circuit and is converted to the binary value 110 corresponding to the base number 6. This value is fed to the microcontroller. The microcontroller runs a program to compare this number with the number corresponding to the destination point punched in at the time of initiating the journey. If they match, the microcontroller instructs the chair to decelerate to a halt. Otherwise, the chair moves forward. Since each base has a unique number identifying it, the IR wavelengths of the corresponding emitters should be uniquely determined by this base number. The UPD-300 IR 1 photodiode from Alphalas GmbH Company or a photodiode of similar capability can be utilized. This photodiode has a response time of less than Ins.
Considering the constant travel speed of the wheelchair to be somewhere around 0.73152 m/s (i.e. approximately 2.4 ft/s or 73 cm/s), and considering the size of the IR emitter to be of a diameter of 0.12 m approximately, the response time of the UPD-300 IR 1 is sufficient to detect the base. Additionally, the processing of the detected signal and corresponding notification for the deceleration to begin takes approximately a few more nanoseconds using high-end MPUs [7, 8]. This means that the chair motor will be almost instantly notified of a halt once it goes over the base, if required.
Once it is detected that the base number matches the destination base number, the motor is cut off and brakes are applied. Considering the wheelchair to be moving at a rather slow speed of about 0.73152 m/s (i.e. 2.4 ft/s), or it will require about 4 seconds and 0.6m (i.e. 2 ft) for the chair to come to a complete stop with a deceleration of 0.18 m/s^sup 2^.
Upon commencement of the journey, the wheelchair moves forward. It acts otherwise only when it encounters an obstruction in its path. Let the wheelchair be negotiating a 90° right turn as shown in Fig. 4. When it was moving forward, it was steadily approaching Wall A. At the same time the sensors in Panels 5 and 6 kept track of the walls on either side of the chair. For the chair to encounter a turn, it must come face to face with a wall like that shown by Wall ?' detected by Obstruction Detector (OD). At the same time there is usually a wall on one of its sides like that shown by Wall B. Here walls A and B form the corner of the bend. The wall on the other side (Wall C) ends at a certain point and bends to the right.
Hence, the wheelchair will negotiate the bend and perform a turn by following the rules as shown in Table 1.
When an obstacle comes in the path of wheelchair movements the obstructions has been detected by means of the three panels of sensors Panels 1, 2 and 3.It is obvious that the wheelchair can avoid the collisions that Panels 2 and 3 can detect but not the ones that are detected only by Panel 1 since it cannot duck under them.
In such a case it comes to a halt and sends a message to the control centre informing of the situation. Additionally, if all the three panels pick up the obstructions simultaneously, then the obstruction may be a wall or a human being standing in front of the chair. In such a case, it will sound a horn and wait for a certain period of time. If the obstruction does not move away (such a case as when the obstruction is a man) the chair notes it to be a wall and moves according to its bend negotiating algorithm. If the obstruction is determined by computation to be of a size navigable by the chair, the chair executes a collision avoidance method  to move past it.
The path-finding algorithm of the wheelchair has to work in conjunction with an effective collision avoidance system in order to negotiate with obstructions on its way . This concept is very appropriate to our perspective and readily applicable in our system. Accordingly, the panels have been made wider than the width of the chair (taking into account the wheels) so that the peripheral sensors can make full use of their position and perform collision avoidance with full efficiency as shown in Fig. 5.
The sensors used for detecting collision are of the Ultrasonic type. The Ultrasonic Range Finder SRF08 is an ultrasonic sensor that has a detection range of 1" - 1 8'. Let the safe distance for the chair to turn or come to a stop be about 4 ft or 1.2 m. Then it has 14ft or 4.27 m to decelerate and travel over before reaching this point. Decelerating at a safe 0.06 m/s2, it will take 11 .7 s to come to a stop.
4. Map Construction and Development of Algorithm
The wheelchair is just provided with the starting point base number and the destination point base number. Using these two pieces of information and the map of the floor of the facility in which it operates, it generates a path between these two points and makes a list of the base numbers in between the starting and destination point base numbers . These numbers are stored in an array and are popped out one by one as the chair travels over them. This method also allows the chair to correct its course if it strays. For that action, the chair keeps a list of the turns it performs in a separate array. When it comes over a base point that it should not have crossed in its path, i.e. it has gone off course, it retraces its path to the last turn and takes the turn in the opposite direction, thereby rectifying its course. The algorithm has been developed considering the certain conditions of the total system. The links between two base points are allocated a particular value (a weight) preferably the distance or the shortest distance between the two points. Each of these values is pre-stored in the memory of the wheelchair's computer. Using these values, the chair is able to calculate the most efficient route between two points as follows:
* A list of links made up by the base points immediately following the source point (present position base point) is made.
* If the destination base point is not one of them, then a sub list is made with the links between the last base points and their immediately following base points.
* This continues until one of the base points in the list is the destination point.
* If the destination point occurs twice in one end list, then the path with the smallest cumulative weight value is chosen.
Let us the map of base points shown in Fig. 6 is considered. Let the source and base points be A and 10 respectively. The path is constructed as follows.
The wheelchair can thus store the base point IDs in the sequence A-B-C and thus keep popping them one after the other as it crosses them and come to a halt once it pops C. The wheelchair will also decide upon which direction to move initially so that it does not travel excessive erroneous distances (i.e. according to the above example, moving right from A, towards D even after constructing the correct path because of it initially facing right from A) by the following method:
The base point sensors located on either walls or on one wall as mentioned before have two parts. These two parts emit two unique signals that the IR receptors in the Panels 5 and 6 of the chair pick up. Owing to the fact that the IR bases are a bit elongated and the command processing take a few nanoseconds with the chair moving very slowly at the onset of the journey, the chair is capable of picking up either of these two signals. The interpretation of the signals by the chair, give it the base IDs of the immediately next base and it can judge whether the course it has set is correct or not. In that way, according to the above example, even if it starts moving towards D initially, once it gets notified of this, it turns around and moves towards C. The Master Algorithm responsible for the automatic locomotion of the wheelchair is as shown in Fig. 7.
5. Discussion and Conclusion
This newly designed motorized wheelchair needs minimal human intervention unlike its predecessors. It does not need any human guidance to find its way from the starting point to the destination set initially by the user. Basic electronic and mechanical component and devices have been used in the making which makes it suitable for large-scale industrial production. Its simplistic design and optimized utilization of all its components effectively makes it a lot cheaper than other similar motor-driven wheelchairs and hence it may be affordable to all. This system has immense leverage for further enhancements and modifications and may be readily implemented inside closed systems like hospitals; rehabilitation centers etc. and hence may be highly useful for the sick people. Detailed research on this system is under progress and only the preliminary results have been reported in the paper.
The authors are thankful to the Techno India Group for their financial assistance in the present research and the Faculty of Agricultural Engineering, BCKV for providing the facilities to carry out this work.
. Linda Fehr, Steven B. Skaar, Guillermo Del Castillo, Computer controlled power wheelchair navigation system, United States Patent, Patent No: US 6842, 692 B2, January 2005, pp. 1-16.
. Robert G. Klimo, Electric wheelchair with improved controlled circuit, United States Patent, Patent No: 4634, 941, January 1987, pp. 1-16.
. I. mirrate, J. Ante lis and J. Minguez, Synchronous EEG brain-actuated wheelchair with automated navigation, in Proc. of the IEEE International Conference on Robotics and Automation, ICRAO9, Kobe, Japan, 2009.
. Raquel Coleman, Multi-feature automated wheelchair, United States Patent, Patent No: 6154, 690, November 2000, pp. 1-15.
. Jerome K. Kurtzberg, John Stephen Lew, Voice controlled motorized wheelchair with sensors and displays, United States Patent, Patent No: 6108, 592, August 2000, pp. 1-6.
. Mohamed Fezari, New Speech processor and ultrasonic sensor based embedded system to improve the control of the motorized wheelchair, Journal of Applied Sciences, 5, 10, 2009, pp. 1750-1755.
. Ervin Weisz, Electric wheelchair drive system, United States Patent, Patent No: 6863, 141 B2, March 2005, pp. 1-14.
. Raquel Coleman, Multi-feature automated wheelchair, United States Patent, Patent No: 6154, 690, November 2000, pp. 1-15.
. Sabyasachi Ghoshray, K. K Yen, A Comprehensive Robot Collision Avoidance Scheme by TwoDimensional Geometric Modeling, Proceedings of the International Conference on Robotics and Automation, 1996, pp. 1087-1093.
.Hiroo Wakaumi, Koichi Nakamura, Takayoshi Matsumara, Development of an automatic wheelchair guided by a magnetic ferrite marker line, Journal of Rehabilitation Research and Development, Vol. 29, No. 1, April 1992, pp. 27-34.
.Eunjeong Ko, Jin Sun Ju, Eun Yi Kim, Nam Seo Goo, An Intelligent Wheelchair to enable mobility of severely disabled and elder people, /CCE", pp. 1-2, 2009, Digest of Technical Papers of International Conference on Consumer Electronics, 2009.
.Raciel Poot, Victor Chi, Lizzie Narváez, Erika Lianes, Maximiliano Canché, Design and Construction of a Prototype of an Electronic Wheelchair Drive for People with Motor Disabilities, IJCSNS International Journal of Computer Science and Network Security, Vol. 10, No. 7, July 2010, pp. 84-90.
.Sho Yokota, Hiroshi Hashimoto, Yasuhiro Ohyama, and Jinhua She, Electric Wheelchair Controlled by Human Body Motion, Journal of Robotics and Mechatronics, Vol. 22, No. 4, 2010, pp. 439-440.
1 Apratim MAJUMDER, 2Niladri BANERJEE, 3Shikha NAYAK and 4B. CHAKRABORTY
1,2Department of Electronics and Communication Engineering
Techno India College of Technology, Kolkata, India
department of computer Science & Technology, Meghnath Saha Institute of Technology
4 Faculty of Agricultural Engineering, BCKV, Kalyani, India.
E-mail: firstname.lastname@example.org, 2ladribanerjeehereíggmail.com, 3nayak.shikhaíggmail.com
Received: 15 February 2011 /Accepted: 17 March 2011 /Published: 29 March 2011