UWB Trilateration with ESP32, DW3000

A project that places three anchors and then calculates the current local location through trilateration on the tag device.
[MaUWB_DW3000 with STM32 AT Command]
github.com/Makerfabs/MaUWB_DW...
[UWB Trilateration with ESP32]
github.com/0015/ThatProject/t...
#UWB #Trilateration #LocalPositioning

[@KuKu427]

Thanks!

[@circleofowls]

It'd be really interesting to build this into a low cost SLAM positioning system. Well done!

[@michaelbeetham2660]

Eric , Thanks for all the ESP32cam information and testing . Tech info is getting hard to find these days. They want us to buy first and regret later.

[@ThatProject]

I agree with you. I also often get results that are different from what I expected. So I do my best to gather as much useful information as possible.

[@BazzzB]

Very nice. This is a building block for room scale positioning. Crucial for smart homes.

[@AndrewTSq]

when do you need this in your smart home? (just asking cause curious :)

[@BazzzB]

@@AndrewTSq Presence on a room/area scale. You can make the lights or music follow you around the house. You can differentiate between multiple people with multiple preferences for each room. This is one of the unsolved smart home issues at this moment. There are multiple solutions, but the position update rate to make it really usable is just not available yet. You need it to be fast enough to turn on the lights in the hallway the second you leave the room, this is hard.

[@AndrewTSq]

@@BazzzB Okey!, thanks :) I saw a interesting sensor for the lights. Its a human presence sensor using radar, and there was someone here on youtube that used it with home assistant, and it turns on the lights instantly. Very cool and they were like $1 each also. It could also tell how many people in the room :)

[@xspeed7434]

Glad to see more on these UWB chips 👍

[@dbagnis]

Very interesting video!!.. Thank you very much for share. I´ll wait for more.

[@JeanPierreLavoie]

Very good experiment and video. You explain well your process and the math behind it. Love it!

[@pouyazarafshan]

Nice job! For 2D RTLS should be fine even you didn't mention if you consider the effect of LoS or NLoS situation in your calculation and since your body block antenna direct coverage and effect in signal quality. For 3D RTLS i would like to recommand you to work on z-axis percision and accuracy in case of indoor installation. Working on network scaling and payload specific usecase can be good idea too. Good luck.

[@prateekmahajan190]

This is awesome, please visualise this in unity in real-time 🎉🎉

[@dalewheat]

A very interesting project and series of experiments. Thank you for sharing this. Also, I find it highly suspicious that you have a two car garage that is empty enough to actually park two cars in it. Something is certainly amiss here. :D

[@aliaksandr5145]

Sensor fusion + calman filtering will work. Just beacons isn’t enough

[@eledikohabib3369]

Have you tried the TDoA? It allows for UpTo 16 anchors or more to be used

[@pouyazarafshan]

@@eledikohabib3369 hello, TDoA is another method to reduce number of anchors in RTLS . mostly 2 anchors sufficient for 2D relative positoning instead of 3. Extending number anchors is related to superframe structure and correct and precise implementation of TDMA in UWB but maximum 16 anchors in 2Way ToF amd 32 anchors in 1 way ToF

[@barth3113]

Great, that's exactly what I was looking for. But I have a few questions: why did u used stm32 with esp32? Can I use only esp32 to do the same (or only with stm32)? If so, which one you recommend?

[@ThatProject]

Unfortunately, there is no library that makes DW3000 work properly on ESP32. If you make a driver, you can drive the UWB module directly on ESP32. Of course, I tried it, but it was not easy to implement this alone at a hobby level. In the end, I gave up and used a product that was easily available on the market. Perhaps the company selling this product implemented a driver for STM32 and made it controllable through ESP32 instead of sharing it, so that it can be hidden. This is my guess.

[@JiyaoZhang-l9v]

Hello, May I ask how to make the clock synchronization? NTP can onl[y reach micro-seconds, but lightspees is so fast, that in meters it will sost only nano-seconds

[@ThatProject]

Quick question. What kind of project do you need something like this?

[@JiyaoZhang-l9v]

@@ThatProject I think mine's quite sam as yours. I'm doing my Bachelor-Graduation project, I shall use LoRa moduls to build up an LPS. I tried to use NTP protokol to apply TDoA method, but it seems quite hard to reach time synchronization. But Signal-Intensity method runs good

[@tienphatnguyen2398]

It is nice video. But can you do the project about IR modules with arduino ESP32S3 ? I kinda struggle it

[@IvanTereh]

Hi, thanks for the video, could you please explore more detailes information about this hardware solution. I interrest in what date update speed are? and what accuracy of this solution (UVB)?

[@pouyazarafshan]

@@IvanTereh Hello , for Decawave DW3000 or DW1000 (now Qorvo mark) uwb technology maximum accuracy in 2D can be around +/- 5 cm and 3rd dimension not accurate than 20cm in best condition. Update rate for each ToF calculation can reach to 160 hz.

[@atabac]

do you detect human breathing? or the human have a fob?

[@ThatProject]

No no no. This is using only UWB tech.

[@atabac]

@@ThatProject i believe uwb can detect human breathing /human presence

[@ThatProject]

@@atabac really? Is there a paper I can refer to for that?

[@AndrewTSq]

@@atabac there are cheap radar modules that can do that atleast. not sure about uwb?

[@atabac]

​@@AndrewTSquwb is rf, why cant it be used as radar? there are many papers using uwb to detect human presence by detecting their breathing.

[@schuylerbrock]

Super cool! How difficult would it be to add a 3rd dimension to this? I'm doing research for a device that will do this and return X,Y,&Z coordinates.

[@ThatProject]

Calculating local coordinates x,y,z using trilateration doesn't seem easy. (I tried, but it was difficult to get reliable values) There was a project I tried in the past using 4 anchors. Check this out. kzbin.info/www/bejne/baDImaN4n9Wombcsi=bhal94qB7I03IH65

[@schuylerbrock]

@@ThatProject Thank you!

[@huckleberryfinn-cz3gd]

Why don't you show (or write) what happens, if you stay still? At least for e.g. 5 seconds. Would be crucial to see how much the jitter or drift is when not moving at all.

[@pouyazarafshan]

@@huckleberryfinn-cz3gdhello, dilution of precision and microcontroller floating point calculation will have major effect in accuracy even tag wad fixed in one location. After that RF characteristics of envirement have second important role in calculation.

[@AndrewTSq]

@@pouyazarafshan but that only applys if they have a clear line of sight? cause otherwise I guess moving objects in the room would interfere with the uwb?

[@pouyazarafshan]

@@AndrewTSq Noise sources that affect the accuracy of Tag position calculation are: multipath effect (exists in LoS and NLoS situations), antenna beam and orientation (these patch antennas have two nulls in pattern related to PCB edges), microprocessor floating point calculation, UWB channel configuration such as percision. signal delay time value inside the firmware, uwb message preamble lenght..., another RF device such as a router in the same frequency band. These sources effect on time of flight calculation regardless of whether the tag is moving or fixed.