05-01-2010, 04:33 PM
[attachment=1090][attachment=1091]
ABSTRACT
Recent evolutionary achievements in robotics and bioengineering have given scientists and engineersâ„¢ great opportunities and challenges to serve humanity. With the development of radar and ultrasonic technologies over the past four decades, when combined with the robotic technology and bioengineering, gave rise to new series of devices, known as electronic travel aids (ETAs). It operates similar to a radar system, sends a laser or an ultrasonic beam, which after striking the object reflects back and is detected by the sensors, and so the corresponding distance from the object is calculated. In particular, these devices are used to help people organ failure and people with disabilities, such as visual impairment, deafness etc. This seminar is about an instrument, which is the outcome of robotics and bioengineering, and it is called NavBelt and the GuideCane. It is a robotics-based obstacle-avoidance system for the blind and visually impaired.
NavBelt is worn by the user like a belt and is equipped with an array of ultrasonic sensors. It provides acoustic signals via a set of stereo earphones that guide the user around obstacles or displays a virtual acoustic panoramic image of the travellerâ„¢s surroundings. One limitation of the NavBelt is that it is exceedingly difficult for the user to comprehend the guidance signals in time to allow fast walking.
A newer device, called GuideCane, effectively overcomes the above problem faced by the use of NavBelt. The GuideCane uses the same mobile robotics technology as the NavBelt but is a wheeled device pushed ahead of the user via an attached cane. When the GuideCane detects an obstacle, it steers around it. The user immediately feels this steering action and can follow the GuideCaneâ„¢s new path easily without any conscious effort.
INTRODUCTION
Recent revolutionary achievements in robotics and bioengineering have given scientists and engineers great opportunities and challenges to serve humanity. This seminar is about NAVBELT AND GUIDECANE, which are two computerised devices based on advanced mobile robotic navigation for obstacle avoidance useful for visually impaired people. This is Bioengineering for people with disabilities.
NavBelt is worn by the user like a belt and is equipped with an array of ultrasonic sensors. It provides acoustic signals via a set of stereo earphones that guide the user around obstacles or displace a virtual acoustic panoramic image of the travellerâ„¢s surroundings. One limitation of the NavBelt is that it is exceedingly difficult for the user to comprehend the guidance signals in time, to allow fast work.
A newer device, called GuideCane, effectively overcomes this problem. The GuideCane uses the same mobile robotics technology as the NavBelt but is a wheeled device pushed ahead of the user via an attached cane. When the Guide Cane detects an obstacle, it steers around it. The user immediately feels this steering action and can follow the Guide Caneâ„¢s new path easily without any conscious effort. The mechanical, electrical and software components, user-machine interface and the prototypes of the two devices are described below.
MOBILE ROBOTICS TECHNOLOGIES FOR THE VISUALLY IMPAIRED.
With the development of radar and ultrasonic technologies over the past four decades, a new series of devices, known as Electronic Travel Aids (ETAâ„¢s), was developed. This seminar introduces two novel ETAâ„¢s that differ from the ETAâ„¢s like C5 laser cane, Mowat sensor, in their ability to not only detect obstacles but also to guide the user around detected obstacles.
Obstacle Avoidance Systems (OAS) originally developed for mobile robots, lend themselves well to incorporation in Electronic Travel Aids for the visually impaired. An OAS for mobile robots typically comprises a set of, ultrasonic or other sensors and the computer algorithm that uses the sensor data to compute the safe path around detected obstacle. One such algorithm is the Vector Field Histogram (VFH).
The VFH method is based on information perceived by an array of ultrasonic sensors (also called Sonars) and a fast statistical analysis of that information. The VFH method builds and continuously upgrades a local map of its immediate surroundings based on recent Sonar data history. The algorithm then computes a momentary steering direction and travel speed and sends this information to the mobile robot. The ultrasonic sensors are controlled by the Error-Eliminating Rapid Ultrasonic Firing (EERUF) method. This method allows Sonars to fire at rates that are five to ten times faster than conventional methods.
In the VHF method, the local map is represented by a two-dimensional (2D) array, called a Histogram Grid. The 2D Histogram Grid is reduced to a one-dimensional Polar Histogram that is constructed around the robot™s momentary location. The Polar Histogram provides an instantaneous 360° panoramic view of the immediate environment, in which elevations suggests the presence of obstacles, and valleys suggests that the corresponding directions are free of obstacles. The Polar Histogram has 72 sectors that are each 5° wide. The numeric values associated with each sector are called Obstacle Density Values. Figure (1), shows the Polar Histogram created from an actual experiment, wherein, high Obstacle Density Values are shown as taller bars in the bar chart-type representation. Hence, the Polar Histogram provides comprehensive information about the environment (with regard to obstacles).
NAV BELT
The NavBelt consists of a belt, a portable computer, and an array of ultrasonic sensors mounted on the front of the belt. Eight ultrasonic sensors, each covering a sector of 15° are mounted on the front pack, providing a total scan range of 120°.The computer processes the signals that arrive from the sensors and applies the robotic obstacle-avoidance algorithms. The acoustic signals are relayed to the user by stereophonic headphones. Figure (2), shows the experimental prototype of the device and pictorial representation of it™s concept.
A binaural feedback system based on internal time difference (i.e. the phase difference between the left and right ears) and amplitude difference (i.e. the difference in amplitude between the two ears) creates a virtual direction (i.e. an impression of directionality of virtual sound sources). The binaural feedback system is used differently in each of the three operational modes.
OPERATIONAL MODES: - The NavBelt is designed for three basic operational modes, each offering a different type of assistance to the user.
Guidance Mode: -
In the guidance mode, the NavBelt only provides the user with the recommended travel speed and direction, generated by the VFH obstacle-avoidance algorithm. In this mode, the system attempts to bring the user to a specified absolute target location. The VFH (Vector Field Histogram) method calculates its recommendation for the momentary travel direction from the polar histogram by searching for sectors with a low obstacle density value. Next, the VFH algorithm searches for the candidate sector that is nearest to the direction of the target and recommends it to the user. The recommended travel speed is determined by the VFH method according to the proximity of the user to the nearest object. The recommended travel speed and direction are relayed to the user by a single stereophonic signal. An important parameter involved in the guidance mode is the rate at which signals are transmitted. When the user is travelling in an unfamiliar environment cluttered with a large number of obstacles, the transmission rate increases and may reach up to 10 signals per second. On the other hand, when travelling in an environment with little or no obstacles, the transmission rate is one signal every three second.
Directional-Guidance Mode: -
In this mode, the traveller uses a joystick or other suitable input devices to define a temporary target direction as follows “ when the joystick is in its neutral position, the system selects a default direction straight ahead of the user no matter which may the user is facing. If the user wishes to turn sideways, he/she deflects the joystick in the desired direction, and a momentary target is selected 5-mt. diagonally ahead of the user in that direction. In case an obstacle is detected, the NavBelt provides the user with relevant information to avoid the obstacle with minimal deviation from the target direction. The recommended travel speed and direction are conveyed to the user through a single stereophonic signal, similar to the method used in the guidance mode. This mode gives the user more control over the global aspects of the navigation task.
Image Mode: -
This mode presents the user with a panoramic virtual acoustic image of the environment. A virtual acoustic image is a stereophonic sound that appears to travel through the user™s head from the right to the left ear. A virtual beam travels from the right side of the user to the left through the sectors covered by the NavBelt™s sonar™s (a range of 120° and 3-mt radius). The binaural feedback system invokes the impression of a virtual sound source moving with the beam from the right to the left ear in what we call a sweep. This is done in several discrete steps, corresponding to the discrete virtual direction steps. Figure (3) shows the graphical representation of the image mode.
At each step, the amplitude of the signal is set proportionally to the distance of the obstacle in that virtual direction. If no obstacles are in a given virtual direction, the virtual sound source is of a low amplitude and barely audible. Otherwise, the amplitude of the virtual sound source is larger. One of the important feature of the image mode is the Acoustic Directional Intensity (ADI), which is directly derived from the polar histogram. The virtual direction of the ADI provides information about the source of the auditory signal in space, indicating the location of an object. The intensity of the signals is proportional to the size of the object and its distance from the person as derived from the polar histogram. The ADI is a combination of the signal duration Ts, the amplitude A, and the pitch.
ADVANTAGES
NavBelt can detect objects as narrow as 10mm.
NavBelt can reliably detect objects with a diameter of 10cm or more, regardless of the travel speed.
The current detection range of the NavBelt is set for 3mt.
DISADVANTAGES
For object with diameter of 10mm, the detection is possible if the objects are stationary or the subject is walking slowly (less than 0.4 m/s).
NavBelt lacked the ability to detect overhanging objects, steps, sidewalks, edges etc. This can be removed by addition of Sonars pointing up and down to detect these types of obstacles.
It does not allow fast-motion.
The NavBelt uses a 2-D representation of the environment. The representation of this type becomes unsafe when travelling near overhanging object or approaching bumps and holes.
The above disadvantage can be removed by substantial modifications to the obstacle-avoidance algorithm and to the auditory interface.
IMPROVEMENTS
The Nav Belt is currently not able to detect over hanging objects. This problem can be removed by using a camera and a laser scanner attached to a special helmet, which can detect objects according to the userâ„¢s head orientation. Adding more sonars to the front pack of the Nav Belt (pointing upwards and downwards) can provide additional information.
GUIDE CANE
It can be thought of as a robotic guide dog. The functional components of the GUIDE CANE are shown in the figure. A servomotor, operating under the control of the built-in computer, can steer the wheels left and right relative to the cane. Both wheels are equipped with encoders to determine their relative position. For obstacle detection, the GuideCane is equipped with ten ultrasonic sensors, and to specify a desired direction of motion, the user operates a mini joystick located at the handle. Based on the user input and the sensor data from its sonarâ„¢s and encoders, the computer decides where to head next and turns the wheels accordingly.
FUNCTIONAL DESCRIPTION
During operation, the user pushes the GuideCane forward with the help of a thumb-operated joystick located near the handle. If the user presses the button forward, the system considers the current direction of travel to be the desired direction. If the user presses the button to the left, the computer adds 90° to the current direction of travel and as soon as this direction is free of obstacles, steers the wheels to the left until the 90° left turn is completed. Functional components are shown.
While travelling, the ultrasonic sensors detect any obstacles in a 120° wide sector ahead of the user. The built-in computer uses the sensor data to instantaneously determine an appropriate direction of travel. If an obstacle blocks, the desired direction of travel the Obstacle Avoidance Algorithm prescribes an alternative direction to circumnavigate the obstacle and then resume in the desired direction.
Once the wheels begin to steer sideways to avoid the obstacles, the user can feel the resulting horizontal rotation of the cane; hence, the traveller changes his/her orientation to align himself/herself with the cane at the nominal angle. Once the obstacle is cleared, the wheels steer back to the original desired direction of travel, although the new line of travel will be offset from the original line of travel. The Guide Cane offers separate solutions for downward and upward steps. Downward steps are detected in a fail-safe manner:- when a downward step is encountered, the wheels of the Guide Cane drop off the edge until the shock-absorbing bottom hits the step “ without a doubt, a signal that the user cannot miss. Because the user walks behind the Guide Cane, he/she has sufficient time to stop. Additional front-facing sonars can detect upward steps. The Guide Cane analyses the environment first and then computes the momentary optimal direction of travel. The bandwidth of information is much smaller and hence easier and safer to follow. Figure (4) also shows the way GuideCane avoids the obstacles.
HARDWARE IMPLEMENTATION
Two basic types of hardware used are: -
a) Mechanical hardware, and,
b) Electronic hardware.
a) Mechanical hardware: -
The Guide Cane must be as compact and lightweight as possible so that user can easily lift it, e.g., for coping with steps, and for access to public transportation. For the same reason, the electronic components should require minimal power in order to minimize the weight of the batteries. The current prototype uses 12AA rechargeable NiMH batteries that power the system for two hours. The estimate of the total weight of a commercially made Guide Cane would be approximately 2.5 kg. Figure (5) shows the mechanical hardware of the GuideCane.
It consists of a housing, a wheelbase and a handle. The housing contains and protects most of the electronic components as shown in the figure. The current prototype is equipped with ten Polaroid ultrasonic sensors that are located around the housing. Eight of the sonars are located in the front in a semicircular fashion with an angular spacing of 15°, thereby covering a 120° sector ahead of the Guide Cane. The other two sonars face directly sideways and are particularly useful for following walls and going through narrow openings, such as doorways. The wheelbase is steered by a small servomotor and supports two unpowered wheels. Two lightweight quadrature encoders mounted to the wheels provide data for odometry. Because the wheels are unpowered, there is much less risk of wheel slippage. The handle serves as the main physical interface between the user and the Guide Cane. The vertical angle of the handle can be adjusted to accommodate user™s of different height. At the level of the user™s hand, a joystick-like pointing device is fixed to the handle. The pointer consists of a mouse button that the user can press with his/her thumb in any direction.
b) Electronic hardware: -
The electronic system architecture of the Guide Cane is shown in the figure. The main brain of the Guide Cane is an embedded PC/104 computer, equipped with a 486 microprocessor clocked at 33MHz. The PC/104 stack consists of four layers. Three of the modules are commercially available, including the motherboard, the Video Graphics Array (VGA) utility module, and a miniature 125-MB hard disk. Figure(5) also shows the electronic hardware.
The fourth module, which is custom built, serves as the main interface between the PC and the sensors (encoders, sonars, and potentiometers) and actuators (main servo and brakes). The main interface executes many time critical tasks, such as firing the sonars at specific times, constantly checking the sonars for an echo, generating Pulse Width Modulation (PWM) signals for the servoâ„¢s, and decoding the encoder data. The fourth module, which performs all these tasks, is called the Microcontroller Interface Board (MCIB). The main interface is connected to the PCâ„¢s bi-directional parallel port. The interface pre-processes most of the sensor data before the data is read by the PC. In addition, all communications are buffered. The pre-processing and buffering not only minimize the communications between the PC and the interface, but also minimize the computational burden on the PC to control the sensors and actuators. The interface consists mainly of three MC68HC11E2 micro controllers, two quadrature decoders, a FIFO buffer and a decoder.
MC68HC11: -
MC68HC11 is a powerful 8-bit data, 16-bit address micro controller from Motorola with an instruction set. The MC68HC11 has in-built EEPROM/OTPROM, RAM, digital I/O, timers, A/D converter, PWM generator and synchronous and asynchronous communications channels. Typical current draw is less than 10mA. Figure (6) shows the connections of MC68HC11.
ARCHITECTURE
The MC68HC11is optimised for low power consumption and high-performance operation at but frequencies up to 4 MHz. The CPU has two 8-bit accumulators (A&B) that cab be concatenated to provide a 16-bit double accumulator (D). Two 16-bit index registers are present (X&Y) to provide indexing to anywhere in the memory map. Although an 8-bit processor, the 68HC11 is a very good processor and some 16-bit instructions (add, subtract, 16*16 divide, 8*8 multiply, shift and rotate). A 16-bit stack pointer is also present, and instructions are provided for stack manipulation. Typically multiplexed address and data bus.
Other features include: -
Powerful bit-manipulation instructions.
Five powerful addressing modes (Immediate, Extended, Indexed, Inherent and Relative).
Power saving STOP and WAIT modes.
Memory-mapped I/O and special functions.
Serial Communications Interface (SCI): -
The SCI features a full duplex Universal Asynchronous Receiver/Transmitter system, using the non-return-to-zero (NRZ) format for Microcontroller-to-PC connections, or to form a serial communications network connecting several widely distributed micro controllers.
Serial Peripheral Interface (SPI): -
The SPI is capable of inter-processor communication in a- multi master system. The SPI also enables synchronous communication between the Microcontroller and peripheral, devices such as: -
Shift registers.
Liquid Crystal Display (LCD) drivers.
Analog to Digital Converters.
Other microprocessors.
Pulse Width Modulation: -
The MC68HC11 Family offers a selection of Pulse Width Modulation (PWM) options to support a variety of applications. Up to six PWM, channels can be selected to create continuous waveforms with programmable rates and software selectable duty cycles from 0 to 100%.
Memory: -
The MC68HC11 Family leads in Microcontroller memory technology. In many applications, the MC68HC11 provides a single chip solution with mask programmed ROM or user-programmable EPROM. The MC68HC11 Familyâ„¢s RAM uses a fully static design and the contents can be preserved during periods of processor inactivity. A 4-channel Direct Memory Access (DMA) unit on some devices permits fast data transfer between two blocks of memory, between registers or between registers and memory.
Timer: -
The industry standard MC68HC11 timer provides flexibility, performance and the ease of use. The system is based on a free-running 16-bit counter with a programmable prescalar, overflow interrupt, and separate function interrupts. It includes additional features like, Input Captures, Output Compares, Real-Time Interrupt, Pulse Accumulator, and Watchdog Function.
A/D Converter: -
A/D systems are available with 8 to 12 channels and 8 and 10-bit resolution. The A/D is software programmable to provide single or continuous conversion modes.
The embedded PC/104 computer provides a convenient development environment. Rechargeable NiMH batteries power the entire system and thus Guide Cane is fully autonomous in terms of power and computational resources. The VGA module is very useful for visual verification and debugging, it is no longer needed after development. In addition, the hard-disk module can be eliminated in the final product because the final software can be stored in an EPROM on the motherboard. For module tests, the PC is connected to a smaller keyboard and a colour LCD screen that is attached to the handle below the developerâ„¢s hand. Figure 7 shows the GuideCane prototype which was extensively tested at the University of Michiganâ„¢s Mobile Robotics Laboratory.
ADVANTAGES
It allows fast walking, up to 1m/s while completing complex manoeuvres through cluttered environments.
It can be used to travel or detect staircases.
Easy to handle, and no extensive training needed.
It rolls on wheels that are in contact with the ground, thus allowing position estimation by odometry.
DISADVANTAGES
It uses ultrasonic sensor-based obstacle avoidance system, which is not sufficiently reliable at detecting all obstacles under all conditions.
It cannot detect overhanging objects like tabletops.
IMPROVEMENTS
The Guide Cane is currently not able to detect tabletops but it can detect these objects with additional upward-looking sonars. The addition of these sonars is expected to improve the Guide Caneâ„¢s performance to a level where a visually impaired person could effectively use the device indoors. Outdoors, however, the implementation of an additional type of sensor will be required to allow the Guide Cane to detect important features, such as sidewalk borderâ„¢s.
CONCLUSION
Both the Nav Belt and the Guide Cane are novel navigation aids designed to help visually impaired users navigate quickly and safely through densely cluttered environments. Both devices use mobile-robotics based obstacle-avoidance technologies to determine in real-time, a safe path for travel and to guide the user along that path. Theoretically, conveying to the user just a single piece of information (i.e. a safe direction to walk in) is efficient, fast, and suitable in practise to full walking speeds and even the image of a particular environment could also be transmitted to the visually impaired person (image mode of Nav Belt). It is fundamentally different from the existing ETAâ„¢s (Electronic Travel Aids) that, at best, only inform the user about the existence and location of obstacles but do not guide the user around them.
BIBLIOGRAPHY
NICHOLAS G.B., SYPROS T., BIO-ENGINEERING FOR PEOPLE WITH DISABILITIES, IEEE JOURNAL, ROBOTICS AND AUTOMATION “ MARCH 2003.
I.ULRICH and J.BORENSTEIN, VFH: LOCAL OBSTACLE AVOIDANCE WITH LOOK AHEAD VERIFICATION, IEEE JOURNAL, ROBOTICS AND AUTOMATION “ AUGUST 2000.
J.BORENSTEIN and Y.KOREN, THE VECTOR FIELD HISTOGRAM- FAST OBSTACLE- AVOIDANCE FOR MOBILE ROBOTS, IEEE JOURNAL, ROBOTICS AND AUTOMATION- JUNE 2000.
CONTENTS
1. INTRODUCTION
2. MOBILE ROBOTICS TECHNOLOGY FOR THE VISUALLY IMPAIRERD
3. NAV BELT: -
OPERATIONAL MODES
ADVANTAGES
DISADVANTAGES
IMPROVEMENTS
4. GUIDE CANE
FUNCTIONAL DESCRIPTION
HARDWARE IMPLEMENTATION
MC68HC11
ADVANTAGES
DISADVANTAGES
IMPROVEMENTS
5. CONCLUSION
6. BIBLIOGRAPHY
