Chapter 8: Conclusions

8.1 Summary

The master objective was to build a portable and affordable personal navigation system for the visually impaired, specifically a system to guide the visually impaired through the streets of a city. This has been achieved during the course of this thesis, through a set of objectives. Each objective and the tasks performed to meet that objective can be stated as follows:
  1. Audible access to map information, and the tools for utilizing this information. This objective was met as described:
  2. The ability to travel outside and know one's position. This objective was met as follows:
  3. An alternative system for the primary guidance system when it fails. This alternative system was designed and implemented as follows:

8.1.1 Results

The results from Chapter 7 clearly show the potential of the ANS. These results surpassed the expectations of the author. As the system currently stands it can accurately track a pedestrian over 500 meters, before the inaccuracy is greater than that of the GPS system. The ANS can be modified and an improved accuracy can be obtained. This improved accuracy can accurately track up to 100 meters, before its inaccuracy is greater than DGPS, and over 2 500 meters with GPS equivalent accuracy.

With DGPS equivalent accuracy for 100 meters, the ANS can be used in more situations than originally designed. Instead of the ANS just tracking the user once the GPS position is lost, the overall accuracy of the entire system can be improved by continuously using the ANS. A number of different scenarios are presented here which improve the overall accuracy and performance of the personal navigation system:

8.2 Claims of Originality

The author started this entire project at Carleton University, September 1992, with a fourth-year project entitled "A Navigational System for the Visually Impaired". [1] This project received a lot of press, which is referenced in Appendix A . As part of the ongoing development of this project, the author worked at Visuaide and at Arkenstone . From the initial conceptualization through proof of concept, design, programming, and testing, the authors contributions were essential to the development of this product. While working at Arkenstone on Atlas Speaks and Strider, a patent was granted to the developers, which includes the author, for this project. The shaded portions in Fig. 13 (p 38) show the specific software modules developed by the author. The design, implementation, de-bugging, and testing of the ANS was entirely done by the author.

8.3 Benefits

With the aid of this Personal Navigation System, the visually impaired can travel outdoors, with the confidence of not getting lost, and the freedom and independence of doing so without relying on others. With the additions of the ANS, the product is complete. The ANS brings reliability and an improved overall accuracy to this system. The cost of adding this system to Strider is negligible, (less than $100).

This Personal Navigation System will track the visually impaired and using synthesized speech present the street address of the current location to the user. Not only can it track the user, but it can also find a route to a specific destination, and then guide the user to this destination.

With "Points of Interest", additional detail can be added to the map making the map more personal by offering custom information. With a "Points of Interest" file open the user can be informed, while travelling, of specific locations as they are approached. For example, if the "Points of Interest" were bus stops, the user could ask for the nearest "Point of Interest", or in this case the nearest bus stop, and be directed to the stop. Upon arrival at the bus stop, the user would be informed.

8.4 Future Improvements

The ANS currently offers only a ten-degree resolution in direction travelled, due to memory constraints. By upgrading the microcontroller to one that has more memory, the two-degree accuracy offered by the compass can be realized. In addition, a separate buffered UART should be incorporated as well as CRC checking and recovery of data.

Due to the limited memory space, the current design does not keep track of the order in which each step was taken and its corresponding direction. For example, if a step were taken north, then east, then north, and finally another step east, either of the following two diagrams could be displayed:

Figure 47: Order of Steps Recorded

In Fig. 47, the figure on the left will be displayed if all four steps are taken before the data is sent to the computer. The figure on the right will be displayed if after each step taken the data is sent to the computer. The user ends up at the same place, but multiple paths could be generated to get the user to this location as shown by the figures. This depends on when the data is sent to the computer. By saving the data in chronological order, the exact path the user travels will always be recorded no matter how frequently the data is sent to the computer.

The integration of the ANS to Strider must also be completed. The time needed to integrate it with Strider is estimated to be less than six months, since the original design incorporated provisions to include it with Strider from the onset. A switching circuit was incorporated so that both the alternative and GPS systems could be connected on the same serial cable. Slight modifications to the device-dependent GPS DLL module will have to be made to receive and signal the ANS to transmit its data. Additional software will have to be written in the "Strider Interface" module (as shown in Fig. 13). This software will convert the steps and directions the ANS reports into a change in the latitude and longitude, which Atlas Speaks uses to display the Strider cursor and the current position solution of the system.

While Strider is tracking the user using the GPS system, the length of the user's stride can be determined "on the fly". This stride length can be averaged over a period of time, and stored. This step-size can be continuously updated, so a current step-size will be used to convert the steps taken into a change in position. Therefore, no manual calibration of the user's step-size will have to be made.

Calibration of the compass's heading information must still be done at least once, so that the compass "knows" where north is. On the prototype board, a switch is used to calibrate the compass. The switch is pressed and the board is turned 180° and it is pressed again. When the compass has not been calibrated it will either give a direction that is an invalid heading or always give the same direction, no matter which direction the compass is facing. In either case software can be designed to recognize this invalid compass state, and to signal Strider that the compass must be calibrated. Strider can inform the user that the compass in the ANS must be calibrated and then proceed to instruct the user on how to calibrate the compass. Instead of physically having to press a switch on the board to calibrate the compass, the calibration process could be done through software, and the user could press a key on the keypad to signal the compass to start its calibration process. Then Strider could tell the user to turn 180° and press the key on the keypad again. This will complete the process, and the user can proceed normally.

There is still one factor preventing this Personal Navigation System from becoming totally accepted and used in the visually impaired community. That is the size and weight of the system, with the laptop computer being the main contributor to this problem. Miniaturization is necessary and an embedded system would be ideal, but the cost of miniaturization may be too much of a financial strain for the visually impaired to bear. One option is to use a palm top computer if it will have enough ports available to support both the speech synthesizer as well as the Strider Box.

Next Previous Table of Contents