OpenStreetMap logo OpenStreetMap

2hu4u's Diary

Recent diary entries

Finished result:

https://www.youtube.com/watch?v=xUkLQj29vE4

Preamble

I have been interested in making “before and after” comparisons of mapping progress for a while. The rich and beautiful OSM-carto style as it appears in standard OSM is a particularly important element of what I wanted to achieve, but it remains quite difficult to render historical map data in this style.

This guide mostly follows the ohsome guide from 2018 but taken further and with updates. Please be aware that I have done little to optimise the workflow so far; this is more of a proof-of-concept at the moment.

I started out mapping in August 2020 and mapped almost exclusively in my old hometown whilst I was stuck there during COVID. I spent a lot of time mapping this area very comprehensively and wanted to visualise my work. Big shoutout to user John Bek who greatly helped with building tracing, which is quite tedious. The Blue Mountains towns of Katoomba, Leura and Wentworth Falls featured are now some of the most comprehensively mapped in Australia.

Previous method

See full entry

Location: Leura, Sydney, Blue Mountains City Council, New South Wales, 2780, Australia

Cheap and easy way to add Bluetooth to Ardusimple simpleRTK2B without XBEE

Posted by 2hu4u on 8 March 2024 in English. Last updated on 4 April 2024.

Introduction

I am writing this tutorial because adding wireless connectivity to the cheapest hobbyist RTK module (Ardusimple simpleRTK2B) is very useful but not trivial for those inexperienced with electronics, and Ardusimple closed down their forums for… some reason. Hopefully this is something anyone can follow to make RTK more accessible to the general hobbyist mapper.

This entry is building on from my previous diary entry “Affordable, High Quality 360 Street Level Imagery using GoPro Fusion and Ardusimple”, I have now been using the rig for a couple years and it is holding up very nicely.

A quick summary is that I used the simpleRTK2B with my phone and GoPro Fusion 360 to make a very cheap and ultra portable street level imagery rig, for Mapillary or Kartaview. It can be used handheld or mounted to a car or bicycle.

The last survey I went on, whilst mostly fine, did bring up some issues regarding the reliability of the USB OTG data connection. This has always been a weak point of the rig in other ways because;

  • the USB connection would drain my phone’s battery and make it difficult to charge the phone simultaneously whilst surveying, unless using wireless charger
  • the phone always needed to be in close proximity to the simpleRTK2B
  • too easy to accidentally disconnect
  • The OTG adapter is wobbly in the type-c port

As my Android app of choice “SW Maps” permits a bluetooth connection for the NTRIP corrections, it was the obvious way to go.

Bluetooth module choice

Ardusimple sells the Bluetooth XBEE module for the simpleRTK2B for €34 which is quite expensive. I know you’re reading this Ardusimple! Third party Bluetooth XBEE modules are somewhat cheaper (around $20 AUD), however I had a generic bluetooth UART module laying around so I decided to just solder it in. Similar ones are quite cheap to buy new, for example the arduino HC-06 module goes for around $10 AUD.

Connecting it up

See full entry

Location: Bentley, City of Canning, Western Australia, 6102, Australia

Affordable, High Quality 360 Street Level Imagery using GoPro Fusion and Ardusimple

Posted by 2hu4u on 31 August 2022 in English. Last updated on 9 March 2024.

Introduction

Taking open-source street-level imagery is a productive and fulfilling way to make a lasting contribution to OpenStreetMap. Typically, amateur street-level imagery uploaded to platforms such as Mapillary and Kartaview consists mainly of images scraped from dashcam footage, or from a handheld smartphone. To a lesser degree, some 360 imagery is also taken. Whilst these are valuable, many street-level imagery sequences suffer from various drawbacks including poor resolution and spatial accuracy, motion blur and narrow field of view. Generally the imagery compares poorly to proprietary alternatives that cannot be used for OSM. In this post I propose a fairly low-budget option for 360 street level imagery that has led to satisfying results. The results are comparible to roughly 1st-2nd generation Google Streetview (circa 2010) and is far better than TomTom/Bing Streetside imagery in Australia.

In this post, GPS refers to geopositioning in general through GNSS, rather than the US satellite constellation.

This post is written from Australian perspective using Android/Windows, so results may vary.

By “affordable” I mean <$1000 AUD. My total expenditure on this project was $700-800 AUD, so that may be inaccessible to some.

I am likely to update this post in the near future as the project continues. Please check back here regularly if you are interested, and feel free to reach out on OSM or the OSM World Discord server.

Camera selection

See full entry

Location: Hurstville, St George, Sydney, Georges River Council, New South Wales, 2220, Australia

Comparison of 360 Cameras on Mapillary

Posted by 2hu4u on 28 August 2022 in English.

The following is an incomplete list of 360 camera sequence examples by model on Mapillary, along with some comments, with respect to the image quality and price of the camera. This will be an ongoing work in progress.

Camera Example Comments
GoPro Fusion 2hu4u Image clarity much worse near stitch line between hemispheres but overall quite good
GoPro Max 360ms; ademturkmen Very similar performance to Fusion
NCTECH LTD iSTAR Pulsar zaf3kala Professional quality 11K spherical images, price unknown but assumed to be very high
LG Electronics LG-R105 yoelt Cheap option, offers OK results but a lot of text illegibile
HUAWEI VOG-L29 lehestener Invalid spherical image?
Insta360 One X2 jkingsleya Image quality doesn’t justify price of camera
Ricoh Theta SC adam2  
MADV/Xiaomi Madventure gness  
Garmin VIRB 360 max93600  

Please feel free to help me add to this list and share your insights using the Google sheets link below. https://docs.google.com/spreadsheets/d/1ajTm_0fRKLHiGJS2TW0CNscbW4NItuyjkJxzU_kbLXg/edit?usp=sharing

The Mapillary GraphQL API can access metadata for camera make/model, however it is an undocumented feature and lately I have been getting “access denied” error even when using my API token. If anyone knows a workaround please comment.