Towards Development of a Robotic Guide Dog

Abstract

According to Statistics Canada, there are 37 million visually impaired people worldwide with about 51,000 in Canada alone. Currently, a guide dog is the primary means of providing mobility and independence for many visually impaired people. Guide dogs assist users in their daily activities such as safely navigating between destinations. Considerable effort and professional training is required to train guide dog beginning when they are only 7 weeks old with socialization training, followed by 5-8 months of training in a Guide Dog Training Centre. The final stage requires the dog and its owner to be trained together as a team at the owner’s residence. This conversion of a dog into a guide dog, not only requires a lot of effort, but is also very expensive with costs exceeding $20,000. This expense is usually borne by organizations such as the CNIB and CCB that often provide these dogs to the blind end user at the cost of $1. However, the owner is fully responsible for the care, feeding and veterinary costs of their guide dog. In addition to the above limitations, a guide dog has an early retirement age of 8 to 10 years; which means that users have to replace their guide dogs throughout their lifetimes. Also, the minimum age requirement to own a guide dog is 16 years in Canada. It is due to these limitations that we seek to develop a cheaper and a more effective solution – The Robotic Guide Dog (RGD). The main idea behind this project is to have a robot that behaves like a real guide dog; with the addition of smart technologies such as GPS, proximity sensors, Internet tracking, stereo vision, and image processing. Currently, our project is in the initial phase where we have divided it into 3 subcomponents: the robot platform, the expansion board and a voice recognition subsystem. We have sourced and assembled a robot which resembles a dog and can be programmed to perform various movements such as sit, walk, and turn left/right. We have also selected and acquired the RoBoard expansion board which will act as a ‘brain’ and provide the ability to connect various sensor subsystems to make the dog more useful in the future. We are also using Sphinx 4, an open source voice recognition package, to recognize the user’s command and perform a suitable robotic function. At the end of this year, we will have completed the first phase of the project: we will fully control the dog from the user’s oral instructions. Future teams will continue to develop this initiative by adding more subsystems to achieve a complete Robotic Guide Dog.