Thanks!

- Checkout Reachy Mini or Stack-chan on orobotio, those are the top 2 SOTA AI personality robots, what you're looking for! Not an exact match, but I see the vision, it's definitely buildable.

- What AI services are you most familiar with? Claude? GPT? OpenRouter? HuggingFace? Gemini? Something else? I'm planning to support all of them, but will be helpful to know which would help you most.

- What would you estimate your skill level with programming, 3d printing, and building robotics electronics? beginner - moderate - expert?

Join my discord, and I'd love to set up a call with you. https://discord.gg/DWfUsCNVs

Thanks for checking it out!

@Jinyibruceli

Hi thanks for the question! HN flagged your comment, so I'll respond here.

You said: "I tried building a hexapod last year and the hardest part wasn't the printing, it was figuring out which servo library worked with which controller board. The Github repos were all over the place with dependencies. Does orobot normalize the BOM and software stack info, or is it mainly aggregating the STLs? I'm curious if the agent-generated content catches stuff like "this uses a deprecated ROS package" or incompatible firmware versions."

1) Yep it normalizes BOMs, whether it's on personal blogs, google docs, github, markdown, html, wherever. Agreed they are all over the place and hard to navigate, but LLM agents are really good at this. As much as it can, it tries to build a single one-click Amazon add to cart with everything in it. If parts aren't on Amazon, it prefers Alibaba next, or they're linked separately. Because some projects online are forks and only have BOM lists for their additions, so you need parts from the fork and the original BOM list, orobot puts both in one spot.

2) Yes orobot writes on-device javascript commands to git pull and run Makefiles or do whatever else it has to do to bootstrap. And that's installed over wifi through the app.

3) Yes the Agent install code definitely should fixup deprecated ROS libraries. Either internally to orobot.io, and it even will post push requests upstream, if there are bugs buried in the source GitHub project. Weather or not it has I'm not sure. Some of this is really new.

I don't want to promise anything I haven't tried myself. Can you link which hexapod github projects you were trying and if they're on orobotio already and what controller board and servos you have? I can try to look up for you how well supported it is right now.

Thanks for the feedback!

There are open source versions of everything done within a GCP API call, but it requires multiple machines and lots of data to build an NLP model to be as fast and accurate as GCP, and cloud computing is relatively new compared to OCR.

There are? Can you give a list of pointers or what to look for?

I was looking for an OCR that can do license plates while the car is moving, for a hobby project. The image quality is less than perfect, the lighting is never very good, and as the camera is mounted on my side window, all plates have a perspective transformation applied (e.g., topline and baseline are essentially never parallel)

Tesseract fails miserably. Trying to help it, I have not found a good open source project that would consistently equalize color pictures to black-and-white - sometimes there's shadow on the plates that foils all simple attempts.

And yet, GCV needs no parameters, and seem to do this perfectly on images I've tried.

So, assuming I'm willing to put in the time - how do I build my own GCV -- even if it's just for the hobby use case of reading license plate (and the next stage: reading house numbers - which GCV does reasonably well, although it is a much much harder problem)

I had some good luck with https://github.com/sergiomsilva/alpr-unconstrained/blob/mast... as long as the images were high enough resolution. You might want to check it out, comes with trained models.

Thanks!

Training the model would be computationally intensive, but deploying that to use Tensorflow.js and predicting a single datapoint in the browser shouldn't be as much, right?

There are ML models that are so computationally intensive that they can't reasonably run on the edge. AI accelerator chips obviously help move the line, but AI accelerators benefit the cloud, too. Furthermore, Models can be tens to hundreds of megabytes in size. Okay for the cloud, not okay for wasm running in the browser.

Also AFAIK GCV uses techniques beyond better OCR that greatly help accuracy. It does image fixing, boundary detection, NLP, spell check, etc.

Isn't this sort of like estimating the value of Ford by paying users not to use roads?

No, the roads are still free - the participants are still free to use other services after deactivating their accounts.

But other services don't connect to your friends. So I guess it's more like, paying them not to use any intersecting public roads.

Is that why Ford decide to stop making sedans? Only trucks can go off-road. ;-)

underrated post.

Wrong, it was the best selling overall sedan in US by revenue, including gas cars.

Have a source for that? Bloomberg says it was the fifth[1] best selling sedan. Also, their numbers include Canada while other manufacturer's do not.

Wait... by _revenue_? Who cares about that? Now units sold isn't the benchmark? C'mon.

[1] https://www.cnet.com/roadshow/news/tesla-model-3-outsells-al...

Investors care about revenue.

Depends on how you calculate it. Toyota sold more car, but in terms of $$ spent, Model 3 is the most popular sedan in US.

No.

OK can you be more helpful and maybe back that assertion up with evidence/data?

Here's some for you: https://www.quora.com/Is-Tesla-losing-money-on-each-Model-3-...

I guess by definition, as long as they lose money, they lose money on each Model 3.

But really what you're implying is, when Fremont is a closer to full produciton, each Model 3 will net a loss. Best engineers say no. https://www.cnet.com/roadshow/news/2018-tesla-model-3-munro-...

Even Tesla's public financials show they lost more money, when they made less model 3s.

The embedded assumption is that current demand is sustainable, rather than simply a reduction in a multi-year fanatic backlog.

Embedded inside that assumption are a lot of assumptions about ramps in production with consistent quality and finish, and lack of EV competition.

I personally disagree that Tesla can maintain production and quality at these levels without needing more financing, but the verdict is definitely not in yet.

You don't care about profitability anymore?

Now you want to talk about demand? OK.

- Tesla is mentioned in multiple current top rap songs.

- Elon has the most viewed episode of the most viewed podcast.

- Tesla is working on a rocket powered car.

- Teslas are featured in major TV shows and movies.

- $45k auto loan is within range of your average middle class budget.

You think that wont translate to sales? We'll see.

I think a lot of promises (and outright lies) have been stated.

https://www.cnbc.com/2018/09/18/tesla-stock-drops-after-comp...

Have the people who complain about javascript page load performance read about netflix.com? Netflix uses JS to render their landing page server side. No client JS necessary. How can a page load faster than raw HTML?

You do realise that Netflix doesn't need to depend-on/load adverts and mass tracking scripts and given the company itself is tech based their sales doesn't get the mighty power to force their devs inserting x40 tracking/adverts scripts and annoying popups? P.S. Netflix does ~17 ajax calls immediately after loading to hydrate its SSRd page ;)