karmicthreat

Unfortunately I am in West Michigan. This is not a place with terribly interesting problems. I probably need to put together my own thing.

Whats the hardest problem anyone here has that they would pay money to have solved?

I can sympathize. I always knew how to program since my teens. I'm 45 now, and didn't really start doing professional programming till I was in my late 30s. I worked hard for about 6 years to build up enough experience to feel confident doing this.

Now I'm at the point where my current job is pretty boring and we don't grow enough to bring in harder problems. But I also really don't want to move out to SV to work on Neuralink, sats, robots. Even though those problems are one that would be super interesting.

How would that even work? Just drop some bacteria that can process inorganics on mars?

I use this for configuration on some equipment I sell. The unit has a local web server that lets you configure IP, ID information. It works as a captive portal. Then you just plug your ethernet cable in and its good to go. (I use a Pi 3 in the product).

Also we have increases the frequencies that we use. Back then it was C-Band, now everyone is using Ku or higher.

Yep! My original team wasn’t quite high enough to be top 9. We merged with another team a member was friends with.

Hack-a-sat is going to be a fun time this weekend. Far enough outside most of our domain knowledge that you have to learn quick and on the fly.

I guess I didn't realize Electric Imp was something people were still using for products. Does make sense for Twilio's API and service lineup though.

It has a linuxy backend with a good gui frontend. And the hardware is usually very good. I just don't really like flapping about trying to get everything working well in linux.

One thing I'm always amazed by when casually talking to people about what I do is how they don't think they can do it. Many people think that software engineering and programming are some inscrutable thing. Sometimes I try to tell them that it is nothing special, it doesn't take a mega genius or math expert to do it. That the most important skill is just being able to solve real problems, soft skills and breaking problems into parts. As well as the ability to self learn and research. I get the idea that most people don't believe that though. Or they think its a humblebrag. Rant done, it's just frustrating that people limit themselves like that.

Not everyone is working on awesome stuff at an intelligence agency. They still need people to work on their 20 year old hellscape of an ERP system in hated language(java,ada,cobol, etc).

All food production is a pretty terrible place to work. The only exceptions might be highly automated foods like diary and egg processing.

That system looks terrible for handling cancer and pre-existing conditions.

My mechanical engineer co-worker loves the depth sensor on the old iPad Pro. He uses it all the time to shapes into solidworks that he then designs parts to fit onto.

I'm sure he will be excited about this if the sensor is that much better.

Hah! This is what originally got me interested in learning to program at 12 or 13.

That system is pretty tiny though. It just covers the main strip downtown. Maybe 6 blocks.

Not really, we were using an IMU with a magnetometer and it was reasonably stable when it was setup correctly. As well as calibrated well. That with the GPS made things stable enough that it was at least a very good demo.

I had considered just some ir lights on the ground and on the mower. Then have a drone hover over the lawn and process the video for positioning.

It’s a good call for iRobot to skip GPS. You can get 1-2cm precision. But not always and it can be hard to tell when you are off. Plus the equipment is still not inexpensive in a consumer product context.

I think if I were to do it over (my lawnmower startup blew up) I’d do something like this and just drive the beacon cost down as much as possible. I had looked at using the DW1000 but the rest of the team wanted something beaconless for commercial applications.

I would trust the beacons a bit more. You could at least test on the mower that you have a good enough beacon constellation.

The ublox zed-f9p is probably the most cost effective right now at 200$

It could work and that plus an IMU is what we were doing at my now recently defunct lawnmower startup. For us to have the precision we needed we had about a 3500$ setup to do so.

That's a pretty neat technique. It seems superficially similar to how mems magnetometers work. The size and oscillation make me question how robust that is though.

One stop gap solution I has was just an IR beacon on my GPS base station. I got an initial alignment with the rover camera and my two gps points. But ended up adding a second gps to the rover.

You would need to get a facing direction. Most robots are doing that with a magnetometer, gyros and an extended Kalman filter. This has quite a few disadvantages though. First and foremost the magnetometer which is really your source of truth is unreliable. It wanders constantly and frequently your Kalman filter will start up in a bad state.

I've seen some multi channel GPS front ends like this one: https://www.crowdsupply.com/amungo-navigation/nut2nt-plus

With 3 or 4 channels you could use the GPS almanac to get satellite direction and your heading. Though you could do it with two if you could get two antennas a couple of feet apart. But smaller robots don't have that kind of room.

Right now I think the biggest pain point with GPS is that if I want single redundancy GPS with 2 CM precision and true heading I need a whole bunch of stuff. 2x 200$ GPS receivers (ublox f9p) on my robot. 1x 200$ receiver with a decent antenna and a xbee radio to act as a RTK base station. And this was just for a simple outdoor rover. And the base station always had to go in the same place because you have to do everything relative to the base station. Then you need all the software to coordinate all that and manage everything.

I want 1 receiver unit, global absolute position within 2 CM with a true heading. Additionally, I want seamless transitioning/augmentation with a global network of terrestrial beacons for interiors and areas with poor surface coverage. And I want all that in a 100$ish module and some sort of subscription.

GPS is one of the most amazing technical onions I've ever dealt with. I've just been scratching the surface of it with understanding positioning for a robotics project. I kind of wish I had an opportunity to do more paid work with it. But it seems to be the realm of PHDs.

One of StarLink's biggest overlooked opportunities I think is radically re-engineering a GPS like system that can work indoors and outdoors as well as offering some pose estimation capability. It would make implementing robots way easier.

I was really hoping for big usability/learning curve improvements with 3.0. I've tried using it and other engineers I know have tried. It just ends up being a complete cluster.

I've read "A Tour of C++" by Stroustrup. It was helpful in trying to catch up to modern idioms for C++.

I don't think that is even a worry for microsoft. Its looking like MS is going to make blazor on electron and eventually blazor on native their cross platform UI.

ORMs have taught me to not rely on ORMs for critical storage. I have one product that uses ruby DataMapper. With my particular product (embedded industrial equipment) it’s terrible. I have to constantly fight it to correct MySQL problems.

If they had just used basic sql it would be much easier to refactor.

Last I looked many of SQL server advanced features didn't work on Linux. https://docs.microsoft.com/en-us/sql/linux/sql-server-linux-...