Collision detection and prevention for the visually impaired using computer vision and machine learning

The white cane is the most common and widely used travel aid for the blind. Its cost is not too high and many government hospitals provide it for free to needy and poor people. It is a great tool that helps blind people detect uneven surfaces, obstacles on the ground, steps, holes, and other hazards. However, it has a major drawback in detecting obstacles that are beyond the range of the white cane, especially when the object is in motion. For example, when a blind person crosses the road, they may not be able to detect a car or person coming from either side (left and right), and it is not much help to the user when navigating through crowded places, which can lead to accidents. Service animals (such as dogs, birds, and miniature horses) are good at handling these types of situations, but they are not easy to train and are very expensive. This paper takes references from many papers and proposes a collision detection and prevention system made of integration of many technologies including image processing, cloud computing, machine learning, IoT and audio production devices, with better option for the blind people for navigating through roads and crowded places and using it along with white cane will make it more effective and will be able to cover some drawbacks of the device.