Cradle of Cantonese cuisine welcomes "robot restaurant complex"
sixthtone.com
sixthtone.com
It was like that scene in The Founder where Ray Kroc eats at McDonalds the first time and he doesn't understand how he could possibly get his burger ready so quickly.
This is easy, just a matter of plating. They probably have a very limited menu (serves 95% just fine) and the rest if just perfecting the system little by little
Burgers and sandwiches should always be hot and ready to pack especially during peak times. Fries too. Others would replenish them as needed. Soda is what it is, so it should a matter of seconds if everyone worked like their livelihood depended on it. But they system is what it is: teenagers making believe they are working and employers making believe they are paying them. In the end, waiting just a few minutes is fast enough for the customer anyway, so no incentive to change the system.
I worked at McDonald’s as a teenager and we really did work. It was easy to get ahead of the game when things weren’t that busy, but often resources were simply over subscribed (not enough workers, too many customers, often both).
https://static.hkej.com/eji/images/2017/11/21/1707152_eb3877...
I always thought that robots and automation would be our hope to compete with the ultra-cheap labor in China. Like Apple moving production back to the US and making it profitable with a lot of automation. But if China now becomes a leader at robotics in addition to remaining competitive at manufacturing, then how are we going to compete?
How was this ever not going to be the case? Robotics will go wherever the manufacturing is. Where else would roboticists get practical real world experience, other than in factories? Obviously there's plenty of robotic manufacturing in America still, but there's clearly no shortage of opportunity in it for China either.
The assumption that America would have the best robotics which would force manufacturing to come back to America seems rooted in little more than wishful thinking.
Industrial robotics on the other hand is a mature field, and much (most?) of it has been outside the US from the beginning. Yaskawa and FANUC in Japan. ABB in Switzerland. Siemens in Germany. How many factories have been automated by these companies? How many have been automated by Boston Dynamics?
Which is not to say advanced tech like computer vision or whatnot don't have industrial applications; it definitely does. Industrial robots using computer vision are already common today, used for things like quality control (e.g. inspecting each individual potato chip on a conveyor belt.) I suspect that to the extent other AI systems might plausibly be relevant to industrial-scale production, industrial robotics companies have been investing. I sincerely doubt Boston Dynamics is going to leapfrog Siemens in any foreseeable future.
Or the one about the art class where one team meticulously plans things out and produces some unfinished shoddy work, while the other team keeps churning out bad pieces but eventually starts churning out good pieces in volume?
Well, at 1.4 billion people, China can most certainly keep churning out bad pieces and readjusting, and they will get much better, much faster than a lot of other countries would.
That it's 2020 and they still aren't seen as the leader of the world in everything has historic roots that have bred a lot of resentment in China towards the West. In particular, the meddling of US in China's affairs.
But the pendulum is swinging. Let's see what happens next.
In that case, the solution might be for governments to massively invest into making the tools available for free to pretty much every citizen, so that in the west, one person can run the same number of experiments as 100 people in China.
Health agencies showing that food preparation ain't a vector:
- https://www.efsa.europa.eu/en/news/coronavirus-no-evidence-f...
- https://www.fda.gov/food/food-safety-during-emergencies/food... (section A worker in my food processing facility/farm has tested positive for COVID-19.)
"Just hours after reimposing the lockdown in Gütersloh, a district with some 360,000 people in the state of North Rhine-Westphalia (NRW), the state officials now reintroduced restrictions in the nearby area of Warendorf and its 278,000 residents.
[...]
after over 1,500 workers at a meat processing plant tested positive"
https://www.dw.com/en/coronavirus-hotspots-flare-in-europe-a...
The venture under discussion Qianxi Robot Catering Group (千玺机器人餐饮集团系) who are owned by real estate group Country Garden (碧桂园 / 02007.HK) who invested 2.4B RMB (USD$33.5M) in the company. It has four subsidiaries including Zhiyuan Technology (智源科技), Bijiashun Kitchen (碧家顺厨), Youbisheng (优碧胜) and Biyouwei (碧有味).
Background is that the parent group sees the combination of robotics, agriculture and real estate as a growth area.
It reportedly spent USD$87M on robotics research in the first half of 2019 and will spent "at least" 80B RMB or USD$11.2B on robotics research in the next 5 years. This includes another company in the group '博智林机器人' (Bright Dream Robotics).
For the current robotics restaurant group, in January 2020 it reports that it will "refine its mass production plan and expand its stores nationwide in 2020."
Includes welcome robots, frying robots, dessert robots, burger robots, bartending robots, claypot rice robots, wok robots, cloud track systems and ground delivery robots, etc.
32 wok based cooking robots and a claypot robot that can make 24 claypot meals at the same time.
My own investors have visited this venture and reported that there were technical problems prohibiting its automatic operation. It seems condiments are supplied by the numerous staff, rather than automatically, and many specific robot interfaces are also handled by staff. For the wok robots, it is reported there are 32 dishes made by robots, and it appears that each wok only makes one (fixed program). Note that the packaging for drinks appears to be the largest and most expensive type of take-away cups available which are presumably selected in order to minimize precision, rigidity and consistency related issues with cheaper plastic lid pressure application processes. There is no evidence of self cleaning. The burger frying robot appears to have the following structure based on two Kuka standard light weight industrial robot arms. The left side focuses on parallel burger frying in sets of three. The right side combines the deep fried (fries, possibly 'fish cakes'). The process is as follows.
A uniform shaped, pre-cut half-bun is grilled and then placed on a new sheet of packaging. Pre-chopped lettuce is added by extracting an apparently weight-measured portion from an apparently hidden rotary dispenser. A chicken burger is fried and then lifted by pneumatic suction. Apparently either the burger or the top of the bun can have sauce applied, or the algorithm has changed. The top is placed on the burger then it is packaged by a complicated folding mechanism. Finally the result is stuck together with a sticker then lifted and placed in the output queue.
Esimate cost of burger system: 2 linear motion systems + 2 robot arms + custom aluminium + grippers + pneumatics + stock framing + heating systems + industrial control systems + air extraction system + frying system + deep frying system + sauce dispensers. Conclusion: expensive, likely over USD$40,000 in parts before labour and design.
Our venture - Infinite Food - retain what we believe to be a global technology lead in the sector and look forward to launching our world first service supplying personalized food retail from a network of kiosk form factor automated restaurants. invest at 8-food dot com if interested.
Anyway, I wonder whether software development for robot can utilize open-source platform, or does it need proprietary software exclusively?
Also, how configurable these robots for different or new menu?
The problem with open source platforms are that they generally provide tooling to target only generic motion systems, and generic motion systems are poorly matched to most specific tasks when it comes to real world metrics. Even if they can perform a task, they tend to be cost, time, space or programming inefficient, require additional complexity (eg. vision system pairing or very high complexity actuators/effectors), come with reliability challenges, etc. Efficient execution relies upon whole domain conception, which includes software, process engineering, mechanical engineering, food safety, cost, reliability, serviceability, manufacturability, IP protection and other considerations.
The domain is simple in conception, but very complex in execution.
The robots by this competitor are not reconfigurable at all, they have one for burgers, one for stirfries, one for basically boiled rice stew, one for ice cream, one for drinks, etc.
We are far more advanced.
Although those robots sure do look expensive. For now. But once you build out the software, the replication costs are minimal. And the robotic machinery will eventually get cheaper, when mass produced.
This video was released just before the virus hit. So it’s perfect for our solitary and isolated post-pandemic world.
Whearas I assume in US, "human touch" is still preferable, e.g. Preferring human operator vs robovoice.
It'll be full of people out the back, faking the experience and constantly repairing robots, for the diners to get the performance.
A great practical robot for a restaurant is a dishwasher.
These robots are straight from Disney.
Instead of small restaurants, you could have ala carte factories cranking dishes under different brands. At scale investment in automation makes more sense.
Right now, most ghost kitchens have one major shortcoming: they lack a compelling narrative. Restaurants don't just sell food -- they sell a story, a distinction. They sell a distinct style of ramen or pizza or pasta because the chef brings their culture and training to the product. Standard markup gets you margins, but a compelling story coupled with good execution adds a premium.
In this regard, pure-play ghost kitchen food feels empty and generic. There's no story or flair behind it, just a factory production line. For the "food as fuel" demographic -- this doesn't matter. Food is food, pizza or pasta is the same anywhere. For people who believe "food is culture" -- this matters a lot. I'm not going to order pizza from a generic Grubhub place named "Sam's Pizza".
One way around this is perhaps to have restaurants with pre-existing narratives to lend their name/brand and process to a ghost kitchen operation. The Filipino chicken chain Jollibees -- which has name recognition -- is trying this out, so we'll see.
For that matter, our world has been "post-pandemic" for longer than recorded history. It's not even the first pandemic in the modern era.
You're probably right, but there's perception. And that's very important for marketing. Pizza delivery companies are advertising their pizza as "untouched by human hand" from the time they emerge from the oven to the time they get to your door.
> It's not even the first pandemic in the modern era.
We have short attention spans. Not everyone is a history expert/pedant like you.
(I’m imagining instant noodles, many baked goods, and sandwiches).
Say they are only targeting to make $1 per meal.
And they’re opened for 14 hours a day, 10 am to 12 am midnight. And say they’re always operating at full capacity. And each table has a 45 minute window between turnovers, so you eat your meal in 45 minutes. So you’ve got about 18 rotations throughout your day. And assume the 600 seats are always optimally filled.
Then, 18 x $600 = $10,800/day
Then, in a year, this could theoretically become: $3.94 million/year.
You could always adjust your numbers, like if they target $2 in profit per meal, then this could theoretically be $7.9 million/year.
Then, at these rates, the initial capital outlay for these robots, might just be worthwhile. Say, you spend $1 million for the robots and machinery, then, it would take 3 months to recover that cost.
Of course, you also have to factor in other costs like the front workers, cleaners, security, and dish washers (unless they have a robot for that too).
But at least you don’t have to pay for a cook anymore. But you will have to pay for a Quality Assurance guy to watch the robots.
And most people prefer to eat during the lunch and dinner rush, so you won’t get to hit the full theoretical operating capacity, thus limiting your profits. But you can try to make more in profit per visitor instead.
I assume with modern sensors, IoT and analytics, it can be done at scale remotely. Instead of maintenance folks regularly inspect on-site, they will only visit when the analytic indicates it is time to fix/clean the machines.
The company can build out 10,000 robotic restaurants like this, which may bring you closer to your estimate of $100M USD per day.
But, there’s also competition. Which will limit your earnings.