3 ms·

So you wrap open source tools/middleware like ROS 1/2, gstreamer, webrtc et al and you want to charge $20/month per robot?! You realize not a single robotic eng

So you wrap open source tools/middleware like ROS 1/2, gstreamer, webrtc et al and you want to charge $20/month per robot?! You realize not a single robotic engineer will use that and they will just use these tools directly for free? It would a good idea to have it open source with none of that subscription crap as mostly in robotics you have either PoC and those will SSH into the SBC and use these tools for free, or a fully funded projects and those will have a team to build a UI around these tools similar to yours, so your audience are slim right here.

chfritz 3y ago
It's not $20 it's $5. And saying that all we did was "wrap" a few open-source libraries is a bit naive. Webrtc is great, but still a maturing technology based on dozens of RFCs that are only in the process of becoming standards. Making it work outside a browser is anything but easy. Making it work reliably over unreliable networks with congestion control, packet loss mitigation, and hardware acceleration, etc. is hard, especially when you need to support OSs that are already a few years old -- as is, of course, common in robotics. If a robotics company will "have a team [to] build" it themselves, will that be free? Far from it! Let's pretend it would only take two people one quarter to implement all this (in reality it took a lot longer -- but maybe we are dumb!), then that's ~$100k to build it and you'd still need to keep paying these engineers to maintain that code every now and then. Even if the company has 100 robots, paying for our service would only cost $6,000 a year. So the $100k would only amortize after 17 years -- ignoring maintenance, which, in fact, may cost more per month than our service. Almost every company I know uses Slack. But why? They could have just implemented a chat themselves, no? After all there are great open-source tools that you just need to "wrap", no? Sure, they could have. But that would've hardly been cheaper, would have born risk, taken time away from their own mission, and the result wouldn't be nearly as good.
amadeuspagel 3y ago
I would say that this reminds me of the classical dropbox comment, except that that comment was friendly advice and this comment is ... not.

chfritz 3y ago

It's not $20 it's $5. And saying that all we did was "wrap" a few open-source libraries is a bit naive. Webrtc is great, but still a maturing technology based on dozens of RFCs that are only in the process of becoming standards. Making it work outside a browser is anything but easy. Making it work reliably over unreliable networks with congestion control, packet loss mitigation, and hardware acceleration, etc. is hard, especially when you need to support OSs that are already a few years old -- as is, of course, common in robotics. If a robotics company will "have a team [to] build" it themselves, will that be free? Far from it! Let's pretend it would only take two people one quarter to implement all this (in reality it took a lot longer -- but maybe we are dumb!), then that's ~$100k to build it and you'd still need to keep paying these engineers to maintain that code every now and then. Even if the company has 100 robots, paying for our service would only cost $6,000 a year. So the $100k would only amortize after 17 years -- ignoring maintenance, which, in fact, may cost more per month than our service. Almost every company I know uses Slack. But why? They could have just implemented a chat themselves, no? After all there are great open-source tools that you just need to "wrap", no? Sure, they could have. But that would've hardly been cheaper, would have born risk, taken time away from their own mission, and the result wouldn't be nearly as good.

amadeuspagel 3y ago

I would say that this reminds me of the classical dropbox comment, except that that comment was friendly advice and this comment is ... not.