I don't know if VBB used this, but most transit agencies publish their data in two standard formats these days GTFS for static schedules and GTFS-real time for real-time data. Any application you build around these formats would immediately scale to pretty much every big city.
Google maps and Apple maps provide transit directions in their apps using GTFS and GTFS real time data (partly the reason why Apple maps was able to add transit directions feature so easily - Google had to deal with the transit agencies years before that and convinced them to publish data in open source standard formats).
It's not like 1password is gonna delete your data or block you from accessing your data if payment fails and your account is frozen. It just gets into read-only mode.
Oh that wasn't my intention - Shazam was and is groundbreaking, they did it when no one else could. All I meant was that it seems more "doable and I probably understand how it works" when compared to how Google assistant recognizes songs from my humming.
Shazam used to wow me, but then as others mentioned in the replies it's essentially matching the signature of the sound to the sounds in the database. If it's one of the song, it gets matched fairly quickly.
It works so well even with my shitty humming - even my girlfriend can't recognize what the song is but Google can. It doesn't even have the same signature as the original audio file, just similar hums in a noisy environment and it still works. Black magic fuckery.
From a quick glance, Notion has the highest valuation per employee (10 Billion USD for 350 employees).
Some of the companies are really bloated imo. Canva has 3100 employeee? Seriously? Yardi lists 8000 employees but no valuation? 8000 seems like a lot, no? That's the same as Stripe, that has a 90 billion USD valuation.
What is the basis of your statement that the factories are never inspected or the products are tested?
I interned at a pharmaceutical company's factory in India for the summer that used to regularly export medications to it's subsidiaries in the western world including United States. The standards are super high, we literally threw away millions of medications due to small uncertainty in a Quality control check.
Besides the local regulations and inspections, a team from FDA regularly visited the factory to audit it for like 10-15 days. Just for one factory. I distinctly remember this as we had a more americanized menu for lunch when the team visited LMAO.
So yeah, when you don't know or are not sure, please don't make statements like "factories are never inspected". It makes it seem like you know something, which you clearly don't.
And what makes you think Amazon doesn't do that yet? Kiva systems did what Ocado does now a DECADE go. Amazon acquired kiva systems (infact, Amazon robotics was born out of this acquisition) long long ago and I'm sure can do a lot more than Ocado does now.
I could be wrong, but I don't think Applied Intuition and Polymath have the same product or business plan. Applied Intuition has a high-fidelity simulator as the main product, Polymath has actual autonomy stack (hardware and software for real world robot) as the product with a low fidelity simulator that gives devs a playground before actually deploying it on a real robot. You can problem train your ML algorithms using the synthetic data from Applied Intuition but Polymath simulator doesn't serve that purpose - they are using real-world data to develop their autonomy stack.
Congratulations on the launch Stefan and Ilia! I'm a PhD student in robotics and autonomy and I've always wondered why something like this already didn't exist, at least for 90% of the 'doable' field robotics tasks. I think you're closer to what Skydio did in the autonomous drones for enterprise space - abstract the autonomy part and just put a little effort into customer-specific requirements. A couple of quick questions:
1. When you say you're built on top of ROS, do you mean the autonomy stack you'd deploy in an actual robot is built on ROS? Are you using ROS or ROS2?
2. What hardware does your current autonomy stack use? For parts of your stack that'd depend on using deep learning based methods (e.g. any image or lidar data), the models that you'd train would have to use a lot of collected and annotated data specific to a particular problem/industry and this would take a non-trivial amount of time, especially since you said the camera/sensor configuration is not fixed can can potentially be decided I the simulator by the user. How do you plan to tackle this?
PS: Any potential internship opportunities at Polymath in the coming few months that I can apply to?
Conda let's you install a specific cuda version directly in a virtual environment with one click though. It's really useful when you have to switch between multiple PyTorch versions and convinient in general imo.
I don't think Jeff cares about SEO. He was out probably looking for a simple domain and I'm sure algorithms.com, algorithms.org etc. were already taken.
If I had to pick between say algorithms.me and algorithms.wtf, I'd definitely go with the second one, especially because students like me will remember it for life.
I was thinking about this the other day: a huge solar storm that can wipe out the power grid in continental United States will also wipe out literally every tech company. Google, Apple, Microsoft, social networking companies, none would exist.
On the other hand, companies like Walmart, Costco, Corteva, etc. would still survive and possibly thrive. Amazon would probably end up opening physical stores.
Yeah, vision is indeed used only to localize, not really to recognize the track. Low-key this is what I don't like Waymo/Cruise approach to autonomous driving - they use HD maps to localize themselves and I see that as finding an easy way out instead of tackling the harder problem.
1. We doesn't have to be another country/military. It could just be another research lab in US, China, or anywhere else in the world. Some people just like pushing tech to the limits, they seek the next hardest problem. The hardest problem post robust drone controls/motion planning era (e.g. drones like DJI) is to make them pure vision based.
2. Millions/Billions? Lol what are you even talking about. It takes about 100k to fund a graduate student for one year. I wouldn't be surprised if this entire research work was less than 1 million, maybe even 500k.
Pretty much every company that tries to capture the environment around it uses a probabilistic model that takes in the raw sensor data and makes predictions about things in the surroundings. The random blinking you see on the Tesla screen is essentially saying that Tesla's model prediction confidence is borderline and changing from instance to instance. Is that great? No, but they can't do much using just vision data. What's the alternative? Show the cars even when your model is not confident anymore just for user experience reasons and give false sense of confidence? That's not ideal either.
Imo I don't know any other consumer car company that shows anything like that on their cars infotainment console. Companies like Waymo and Cruise do a much better job at capturing surroundings but they are in a different market where it makes sense to spend a ton of money on each car to add many LiDARs and significant compute in the car.
What do you mean? In a flight simulator you have your controls controlling a plane (right body, dynamics are well known) in the sky (simple environment, dynamics are reasonably well known, even with different atmospheric conditions like wind). The output is visualized on the screens as a rendering of the scene.
In the case of robotic surgery simulator, you are using the controls to control the arm, but to interact with what? If you just want to move the arm around and maybe interact with some rigid objects sure that's easy. Would that add any training value to the surgeon? Probably not. You can get value only when the simulator includes a simulation of something the surgeon would have to face eventually - organic mass of the human body. Simulating that is hard, and I doubt anyone would invest much into it when you can train surgeons on alternative physical objects like pigs.
The worst part is they were never really transparent about what the issue was.