Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Best technologies for indoor position tracking?
23 points by benjaminogles 13 days ago | hide | past | web | favorite | 11 comments
I have a hobby project in mind but I'm not sure which technologies will give me the most bang for my buck long term.

I want to track my 3D position accurately within my apartment. My first use case is to track how long I sleep and how long I watch TV etc. without using some sort of reporting mechanism. I would like to incrementally add more advanced features such as gesture recognition for integration with smart devices etc.

I have read about using RF signals and signal processing to track 3D motion indoors without using wearables (https://www.emeraldinno.com/publications). There must be ways to do it with wearables and bluetooth. Posyx (https://www.pozyx.io/) works with UW signals.

Has anyone had experience with a similar project? Which technologies did you use? Is there specific hardware you would reccommend?


I used to work for an IoT startup on Boston that worked on this exact thing and then the company folded four days after I got married, two years ago. Fun times.

Anyhoo, we used Impinj readers and RFID tags to track things indoors. One of our clients was a Norwegian shipyard who wanted to keep track of their tools because apparently, tools for fixing ships are expensive and people like stealing.

You put the Impinj readers on the ceiling on both sides of a doorway and make sure your API throttles the reads so you can make sense of the data deluge.

I just wrote/designed the dashboards and interfaced with our internal API. If you have more questions I can ping the guys that worked on the back-end.

Does it work for accurate positioning though?

Many thanks for your question! It prompted me to rethink a problem in Home Telehealth and Elder-Care having to do with tracking Dementia patients. Anyway, a PubMed paper published 2018-1-24 titled: Passive Infrared (PIR)-Based Indoor Position Tracking for Smart Homes Using Accessibility Maps and A-Star Algorithm (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5855945/) may help you.

The paper is about using passive infrared sensors (active badge etc.) to accomplish indoor position tracking. This may be overkill for your need but I figured I'd put it out there for you anyway. I hope to have some follow up information in a couple of days and will revisit this question.

I tried to do this, I wanted to track my cats. My problem was that I was trying to fix this by using one tech. I did a fair bit of research on Bluetooth, cheap 433mhz radio, accelerometers etc. All had drawbacks, from size, cost, availability, missing features (eg the signal strenght) and I gave up when I moved to London.

If I try this again, and by that I mean when I try this again, I will go a different route. The core will be a pair of accelerometers, with a nfc tag, sending to a central unit the acceleration data and wifi signal strenght. A network of nfc readers will send info about what tags are they seeing. Maybe even some cameras.

The acceleration data will provide fine movement data and the rest is supposed to allow for data calibration and coarse tracking. Not an easy project, but should be fun.

If you don't mind me asking why would you want to do this? We are working on technology for a "digital twin" and have not really thought about applying it to inside the home. What benefits are there of knowing where you/other people are around the home?

Good question, my main motivation is to learn more about how such a system would work. I am interested in signal processing and especially radar like systems so implementing one would be a fun experience.

In general, systems based on radar are trying to replace wearables for sleep states tracking, elderly fall detection etc. But I'm not sure yet what other applications it might have. The ability to track my location just seemed like a good first step.

If you don't mind wearing a Hololens all day, it orients itself by turning your entire surroundings into ~1m mesh blocks that serve as its internal map. As a bonus, you can retrieve the current mesh from the device's internal web server in order to visualize against it.

Cons: Wearing Hololens all day, spending thousands on a hobby project. :-/ Fascinating if you manage to borrow one though.

A smaller and less expensive V-SLAM option would be the Intel RealSense T265 camera.

Tracking accuracy can sometimes drift quite a bit with this if you’re not careful, but it can perform pretty well, especially if loop closing is enabled.

> Fascinating if you manage to borrow one though

Even more fascinating, if they manage to wear it the whole day.

Thanks for the info! We are actually approaching it from a more computer vision angle. Think several RGBD sensors (kinect/structure) placed around your home and fusing the data into a real-time 3D model.

Interesting approach using radar for tracking, if you make any progress would be interested to hear about it.

A small set of well-placed cameras, image processing, and some number crunching. You know... draw the rest of the owl ;)

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact