Improving Indoor Navigation Of Robots With IR

If the booths at CES are to be believed, the future is full of home robots: everything from humanoid robots on wheels to Alexas duct taped to a Roomba. Back in reality, home robots really aren’t a thing yet. There’s an obvious reason for this: getting around a house is hard. A robot might actually need legs to get up and down stairs, and GPS simply doesn’t exist indoors, at least to the accuracy needed. How on Earth does a robot even navigate indoors?

This project for the Hackaday Prize solves the problem of indoor navigation, and it does it in an amazingly clever way. This is using QR codes for navigation, but not just any QR codes. They’re QR codes read by an infrared camera, and painted on the walls and ceilings with a special IR sensitive paint that’s invisible to the human eye. It’s navigation for robotic vision, and it’s a fantastic idea.

The basic idea behind this project is to use an IR camera — or basically any webcam with the IR blocking filter removed — and a massive amount of IR LEDs to illuminate any target. So far, the proof of concept works. A computer can easily read QR codes, and if paint is invisible to the human eye but visible to an IR camera, the entire project is merely a matter of implementation.

There have been a number of projects that try to add indoor navigation to robots. Some of them use LIDAR, some use computer vision and SLAM. These are computationally expensive. Some even use wireless beacons to navigate indoors like the SubPos Ranger from the 2016 Hackaday Prize. Using IR and QR codes is just so simple and hacker-friendly, and we think it’s fantastic.

20 thoughts on “Improving Indoor Navigation Of Robots With IR

    1. Hello Andrew , this is my project. Printers and Print cartidges can be modified. So far the easiest solution I have found is using an epson printer with 702xl cartridges, I purchased empty clear cartridges and mixed in inks and filled them. Also if you dont want to expense there is a simple method of purchasing an empty refillable marker and filling it with IR upshift ink or UV invisible ink. Then you can draw a qr code. Surprisingly even drawn qr codes work great.

    1. Yes you could put QR codes outside, but you will need to choose the ink either upshift or down shift based on the color of the object being painted. and other anti counterfeit companies produce pigments and inks for this.

      1. I have north star beacons to use in comparison, I can tell you there is an advantage of this. First north star beacons are usually a IR led pointed at the ceiling. It doesnt work well if the ceiling is not flat, angled ceilings really complicate things. Second is that daylight can prevent your robot from seeing a reliable set of dots to triangulate on. Third is that this only gives a method of localization , but not identifiers like which direction the robot is pointed, what room, or other information you may want to embed in the QR code. your right, they are more than just a pain, they cost a lot too.

  1. Interesting the QR codes could of course explicitly convey location data (this QR code is known to be printed on the bathroom door for instance) but if the orientation and scale of each QR code is recorded along with its location then any transformation done in the process of reading the QR code (using its normal built-in registration mechanism) could yield clues to distance, angle, and elevation as well. This is super nifty.

  2. Hmm, your post pushed my mind to a chain of thoughts, about what makes a graphical shape useful for orientation, and at the end of it, it came to my mind that road navigation for autonomous vehicles is done wrong right now because visual clues, signalization, is too ambiguous, information-poor, and with too little redundancy for computer systems. Road marks and signs, as well as vehicle marks (license plates, but also lights), have to be designed with features both meaningful to human drivers and useful to robotic ones. Motorisation of road traffic was not outperforming much (or at all) the horse carriage on the roads of their time, and couldn’t even compare with trains of the day in speed, reliability and safety, and automation of road traffic will also need introduction of significant changes to the infrastructure before it becomes safer, and therefore useful.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.