Edge Based Obstacle Detection Model for Outdoor Type Obstacles

Obstacle detection and avoidance technologies are mainly categorised into non-vision and vision-based technologies. Most of the developed technologies are not ready for public use as they may require additional stage such as production and distribution. Obstacle detection model built for a mobile device focusing on detecting outdoor obstacles is introduced in this paper. The model uses a monocular camera to obtain real-time frames then applies Canny Edge detection algorithm to obtain the surrounding information. This surrounding edge information is used to compare between frames to determine whether a path contains obstacles. In addition, the proximity light sensor, accelerometer, and gyroscope are used to ensure the model’s adaptability to various environments. The model is tested in various out door scenarios. The average floor-based outdoor obstacle detection accuracy obtained by this model is 81.7%. This concludes that the model can be a supplementary assistance to the white cane, which is used by people with low vision.