we are getting ready to install hardware at a new location for work (at Pipedream Labs) and didn't want to pay $4k for a lidar map from a survey company, so I decided to fix a broken sensor I bought last week for $200. read along below to see... what needed to be done.show more

I was just planning to fly a drone and build a map for CAD purposes using photogrammetry, but I'm always disappointed with the ground-level details using that method. I just bought this old Velodyne VLP-32C, but it is in rough shape, had a cut cable, and no interface box.

pulling it apart, things are looking pretty good. no obvious binding in the spinning mechanism, optics look intact, and no signs of anything catching on fire. I don't know who cut the interface cable this god damn short, but they should feel bad about themselves.

there's a molex connector inside, but getting to it requires removing the spinning assembly & pulling out glued hardware. I'm splicing into the existing (short) cable first to see if the unit is functional, and then coming back to tie directly into the interior connector later.

had to find a used interface box on ebay -- these work for most of the 16 & 32 sensors. Pulled it apart, then got down to business stripping the cables on both ends. I used an exacto knife and went real slow to be sure I didn't lose any wire, margin for error is essentially zero.

stripping, tinning, and soldering these together is a nightmare. I suck at soldering but I think I would've hated this even if I was good at it. fuck short wires.

first spin-up! pulled a power supply after checking the hardware manual and plugged it in. hell yea.

velodynes can be configured to route their packets directly to a specific IP. Plugging the ethernet connection into my laptop and pulling up wireshark confirms this, with the sensor at 192.168.10.201. setup my ip/subnet and was then able to pull up the web interface. great news.

everything seems happy and packets are now flowing to my laptop (confirmed with wireshark)! I generally route these things into @rosorg drivers to get point clouds for robots, but velodyne has a nice debug app that runs on osx. here's me waving in my workshop.

next step is collecting sensor data to generate a georeferenced map. I generally use these maps for localization & perception, but we just need it to plan our dig and pipe/hardware install. if youre interested in the map creation process, give me a follow -- that's next.

I recorded a 31-minute video on how 7-8 figure eCom brands can add $143k using 3 Klaviyo "tweaks". Watch the exact method we used recently with a client. Click below.

Very cool. What type of site are you surveying (indoor, outdoor, underground…), how large, and what precision do you need? We’ve lidar mapped 200+ cities at the ground level. Happy to share some things we’ve learned along the way or point to a cheaper off the shelf solution. SLAM can be a bitch depending on the environment. High end IMU tightly time synced with lidar + intrinsics helps a ton

@pipedream_labs I’ll probably go the slam route since I’ve been doing that by hand for ten years. LIDAR only algorithms are pretty good now, depending on the physical structure of the site. This one is outdoors!