Sensor for tracking relative position of human

Question

I am making a robot that needs to continuously track the relative position of a human, up to 15 meters away and with at least 300 degrees coverage. Currently I am using a Hitechnic IRSeeker v2 sensor and made a beacon wristband with 6 TV remote IR LEDs. But the maximum distance I can get is around 3 meters.

I ordered some 3 watts IR LEDs to boost the power, but the size of the wristband will be a problem because it will not run on a CR2032 battery.

I also bought some IR remote receivers. But I am not sure if the reflection from the wall will give false results.

Is what I am trying to do possible? Is a beacon 15m away feasible using this technology?

• If it is, then what do I need to modify in my current implementation?
• If not, are there any other technologies that I should be considering to track the relative position or direction of human, 15 meters away with at least 300 degrees coverage?

Answer 1

There are two environments namely indoor and outdoor environments. If it is indoor, then there may be interference issues like what you have mentioned - walls. In the outdoor environments, (particularly during sunny days) cheap IR sensors can encounter false-detections and that has to be compensated by using IR filters (hardware method) or using digital filters (software method; unreliable) and they are not directional generally.

"Triangulation" is one of the most popular techniques and three WiFi/similar hotspots along with a mobile WiFi/Bluetooth device like embedding it in a band that you have mentioned are needed. However, if obstructions like walls occur, then, it is wiser to use electromagnetic rays that penetrate the walls (there will be many factors involved in the penetration of rays through concrete like thickness, composition, paint etc) and this is an active research topic. One of the easiest methods is to use several cameras at multiple locations and use cascade object classification and/or use a combination of triangulation.

Answer 2

A rather inexpensive solution to your problem is to use a vision based IR solution- Infragram. There is some good work done to post-process gathered IR images and you could post process you "human" images and tune them for your application. It is cheap and light which is why I used it for my open-source projects.

Also, addressing Prasad's concern, this works in both indoors and outdoors environments. Alternatively, if you really want to be adventurous, take out the infrared filter of any camera that you have and convert it into a low-cost IR camera that can detect human heat signatures (Beware: this might render your camera useless and is generally difficult to do. I've tried it, but damaged the filter beyond repair)