Detection of Stairs using Ultrasonic Sensor - Using just sensors

Question

I am working on a project for blind navigation wherein Im planning to place ultrasonic sensors like two each on the shoe, four on the vest( the person will be wearing) and one on his/her cap. In this scenario, how can I detect upstairs and downstairs. Any idea.

I read quite a few research which specifically address this issue, but not gives appropriate answer. Like most of them are not practically for me.

Answer 1

Normal Ultrasonic sensors have 60 deg. Radiation pattern so regularly they can detect everything in their line-of-sight. When someone wear those shoes, he can detect the upstairs as an obstacle, but he can't detect downstairs because ultrasonic sensors wide beam can't tell you that is an upstairs or downstairs. You can set two pairs of ultrasonic sensors in each shoe (four for a pairs of them) one of each shoe must look upward and another must look downward, but this way has it's challenges and i think it's not gonna be useful so much.

I know that you said you are going to use Ultrasonic sensor for your project but according to the reason that i mentioned it it's not gonna be successful or at least it's going to be extremely hard challenge( and i tried to find a solution at every possible ways that i thought about it, but i didn't make it)

I think for this purpose you have two options :

1. somehow you narrow the beam of your Ultrasonic sensor using a parabolic antenna and install your sensor as it's feeder part of it (of your parabolic antenna); and scan your environment by rotating the antenna in suitable speed and angle in 2-axis; but because of ultrasonic-sensor's limited range, it's not work for a person and it works for a small robot. But using microwave obstacle detector and doing this technique to narrow it's radiation pattern and follow that those instructions have to make a good results.

2. using environmental scanning LIDAR (laser radar ) is extremely accurate and useful.

If you didn't choose these 2 options, you have another two below options :

• usind SAR (Synthetic Aperture Radar) that is using for Geo-mapping and environmental scanning and you can find it in MIT University Open Course-Ware by searchings "MIT Open Course-ware Synthetic Aperture Radar" through the internet

• Using AI-powered vision system by a camera and suitable processor in purpose of obstacle detection and you have to use Convolutional Neural Networks to terrain your artificial neural networks at your PC and upload your trained Artificial Neural Network on your Processor. it's best option ( the most precious and cheapest solution ) that i know, because automatic drive mode in Tesla machines using the same technique (something like this) and if you have no experience in AI, don't afraid, everything is easy if you really love to do it.