Lidar-equipped, 30-pound robot dog can be yours for $1,600 | Hacker News Reader

Lidar-equipped, 30-pound robot dog can be yours for $1,600

arstechnica.com

43 pointsBrajeshwar27 comments

If they can stick a LiDAR on this for $1.6k it kind of negates Teslas argument that LiDAR is too expensive for consumer cars. That or it's not really LiDAR on this.

iPhones have LIDAR, so you could already make that case. I doubt that all LIDAR is the same, though, and there are more expensive solutions for increased range and speed, reliability, durability.

Tesla's was more likely motivated by supply chain issues, though, as well as what seems to be a weirdly motivated Elon sticking point that humans can drive successfully with just vision.

There are different kinds of lidars that use different wavelengths and sensing methods.

For self driving car use, the difference between a small lidar that's on a phone or vacuum cleaner and the one on a car would be the amount of power used such that objects far away can be detected, and the ability to operate in automotive environments (noisy power supply, high temperatures, lots of vibration).

Tesla’s argument centered around the idea that vision and LiDAR were not complementary. They questioned the reliability of having two different sensors providing conflicting information and the challenge of determining which one is accurate.

So kind of like breaking one of two compasses before going out to sea. Makes perfect sense.

Web developers building cars have yet to learn about Kalman filters and basic signal processing.

Am I taking crazy pills? How does this make perfect sense? The obvious solution is to have more compasses and redundancy.

They don't provide conflicting information. They provide complementary information.

I guess that dog's LiDAR doesn't need to read much longer distances with maximum accuracy at like 150+ Km/h speed; then add certifications et al for automotive use and you probably have very different costs.

When tesla was first saying that LIDAR was nothing more than a laser and receiving optics mounted on a spinning hub that spun real fast. Like old those big radar dishes on big navy ships.

In the years since, ic and component manufacturers have created solid state lidar. They're basically little electronics modules that cost a few dozen dollars, no motors, etc.

The world changes. Better components come along

What are some small and lightweight (compatibly) LiDAR on a chip that you’d suggest looking into?

There are lots, here is one: https://www.analog.com/en/applications/technology/lidar.html...

The video looks completely CGI and so I doubt any of the rendered examples will actually happen.

I need to see this dog operate in a real video to determine whats marketing BS vs reality.

I disagree. (By CGI you mean, fake reality, such as superman flying through air, right?)

* I saw "3d mesh" overlay which was CGI, at 1:06 https://youtu.be/6zPvT0ig1VM?t=66

Everything else, showing the robot moving up/down stairs, jumping, looked like reality to my eyes.

What other parts of video appear CGI?

Also, look through their other YT videos: https://www.youtube.com/@unitreerobotics/videos they have published a lot of video, also not looking like CGI.

Compare an older model, to SPOT from Boston Dynamics, four years ago: https://www.youtube.com/watch?v=rytwDPIYZ20

EDIT: 11 months ago, a review of the older G1 model https://www.youtube.com/watch?v=09W8EWIZl1k - seems to be real

The people don’t look real to me.

The shadows and textures on the dog jumping don’t look.

Even the backgrounds, grass, walls, buildings don’t look real.

Is this supposed to be actual video of a location with a real human jogging alongside the dog?

Ugh, idk, especially the part about moving around (running besides an animal, moving in the grass in the ISS section) were too clean. Like no dust, no specs of dirt from the grass nor any kind of traces of actually having touched anything.

The person with the green shirt also looks either like hq CGI or a heavily edited recording where the dog was CGI'ed inside. The shadows as OC pointed out were in some cases also too distinct though mayve this is due to artifical lighting.

Overall I could see the bot being real but that demo video sure ain't a true depicition of reality.

How much with the arm though? I would only like something that has an arm so it can potentially be programmed to do something useful. Maybe many versions away that will become actually useful but still, working towards that seems more interesting than having something basically worse at everything than my actual dogs.

I'd be more interested in one with wheels I could ride on. Like a smart scooter that drives itself.

I'd be very creeped if I saw one of these in real life, but I can imagine them being used for security. While pay a security guard when Robo dog can just as easily walk around and call for help if needed.

Why buy a 1600 dog when you can buy 16-32 cameras and cover the area?

Interesting, and I'd be seriously tempted, but I'm concerned it requires to be connected and phone home to work, which in my house is a no-no for anything that is not Open Source, especially if it's equipped with cameras and can wander around.

I always wonder if this is the economy of scale or they sell these at a loss. I think just buying off-the-shelf bldc controllers would cost me more than $1600.

I’m always looking at products coming out of China to see if they would benefit the Chinese government in some way. Selling items at a loss wouldn’t matter to them if the data received is worth more.

This product is constantly scanning LiDAR data, would crowd sourced LiDAR data benefit them in some way? Constantly-on roaming camera and microphone to supplement the HikVision Camera spy network? Mobile Bluetooth scanner for airtag-like target tracking?

Maybe I’m just paranoid but that’s always my thought process.

I apply that to all consumer products and switch China with « NSA »

So do I, but I don’t worry about America trying to take over the world in the same way.

Y’all already did in many way.

That’s fine, I would rather have a US master.

But that soft power bit worked wonder. It’s kinda stunning in the economics and social aspects of life.

For instance: I think we might worse off to address climate change thougt the usa-provided glasses of capitalism.

The data stream for that would be staggering.

Sure, we’re technical so we know some limitations. But I’m sure you know I don’t mean a continuous stream of everything all the time, the same way Tesla isn’t crowdsourcing by streaming all data all the time.

Reply on news.ycombinator.com