Berkeley SkyDeck Application Outlines
This page lists key considerations for the application to become a SkyDeck startup.
Introduction
According to a reliable internal source, Berkeley SkyDeck will start accepting applications for startups on April 2, 2019. I see this as an opportunity to advertise the robotic restaurant model that we have spent much time working on. In this page, I will list some considerations for us to join this competition.
Statement of Purpose
The robotic restaurant is a new operating model for the food service industry, integrating trending technologies such as Internet of Things (IoT), intelligent robotics, QR code, and visual imitation learning. While this model seeks to improve efficiency across restaurants, it doesn't necessarily obviate the importance of human intervention. Quite contrary to the prevalent fear that intelligent machines would one day take over human job, the robotic restaurant model will add new and interesting roles for human employees. In other words, the nature of job will change for human, but not its very existence.
At this stage, we are not planning to open a new restaurant, but develop a smart solution for millions of restaurants around the world to keep them up-to-date with latest technologies and bring to them highest standards of efficiency.
Technical Underpinnings
Customer Experience and Back-of-House Operations
When a customer enters the restaurant, there will be a tall machine at reception. This machine takes in a phone number and quickly searches for available tables. When it finds one, a text message will be sent to the customer's phone indicating the table number and seating capacity. If a queue is ahead of the customer, this machine will send a text message that includes the queueing status.
After settling into seats, customers can place an order through an online app. Instead of having to download one from the App Store, they will scan a QR code on the table. And they will be prompted to enter a mobile application where they can place an order. By accessing the application at any time, customers can view the status of their orders. If notifications are enabled on the phone, customers can also get alerts when orders are made, meal cooked, dishes delivered and check paid.
Once an order is placed, it is assigned to the immediate next available robotic chef. There are multiple robotic chefs around the kitchen, each capable of handling one order at a time. Back in the kitchen, the inventory takes the form of a vending machine mechanism. All the raw materials are stored in this compartmentalized machine. When a robotic chef sends request to the inventory, the type and amount of ingredients will be read and released. Upon dropping each load of raw materials, the inventory pushes the load onto an electrified conveyor system that delivers the load to the robotic chef that calls it.
Each robotic chef is itself a functional unit, having close to it a range of condiments and cooking utensils. A glass screen separates the unit from outside, providing a glimpse for customers who wish to observe how robotic chefs work. The glass is closed when a chef is working, and open otherwise.
There is a fleet of robots around the restaurants, serving as liaisons between tables and the kitchen. When any robotic chef gives an alert that a dish has been cooked, the immediate next available robot will respond to the call and transport the dish across the restaurant floor.
Robots are also servants that provide various forms of customer support. For instance, a robot will respond to a table request for water fills.
When customers leave, they will complete payment on the same system where they first place the order. Like their entry into the restaurant, no formal farewell will be given. However, customers will receive an email soon afterwards that details their expenditure, a brief thank-you message, and possibly some targeted promotions based on their menu choices.
Key Components Description
Central System
Should be fully enabled with Wi-Fi connection, Internet of Things (IoT) technology and data analytics tools.
Robots
Each robot can respond to meal alerts from the kitchen, and takes the meal across the restaurant floor to reach individual tables. When this action is under way, the Central System should bundle together the robot, the table, the order and the robotic kitchen. The robot in action should not be released from the rest three objects until meal is delivered.
Each robot can also respond to customer demands such as water fills. When such actions are under way, the Central System should bundle together the robot and the table. And the robot will not be released from the table object until the current demand is fulfilled.
Each robot has two statuses: busy and available. Under busy mode, the Central System removes the robot from the list of all available robot objects and places it into the unavailable list. When a task is finished, the robot is transferred from the unavailable list back into the available list.
Each robot should be equipped with a navigational system that allows it to trip across the restaurant floor. It needs to know where to turn and stop for passers-by. If necessary, we should paint specially designed routes on the restaurant floor and have a robot object stick to it.
Each robot should possess machine vision capabilities, which through object detection enable it to distinguish moving objects such as passers-by from stationary objects such as dishes on a table.
Robotic Kitchen
When there is an order object in the Central System, it matches an available robotic kitchen to prepare it. When such action is under way, the selected kitchen is bound to the order and marked unavailable.
Each robotic kitchen has two statuses: busy and available. Under busy mode, the Central System removes the kitchen from the list of all available kitchen objects and places it into the unavailable list. When a meal is completed, the kitchen is transferred from the unavailable list back into the available list.
Inventory
An inventory stores raw materials based on their different categories. It receives signals from the Central System about the type and amount of ingredients to release for each order.
When dispensing items, these items fall into a container at the very bottom of the inventory through a funnel-shaped channel.
Conveyor System
Container pushed from inventory onto the sortation conveyor.
Different containers fall into different gravity conveyors mounted at a slight decline, each leading to a different robotic kitchen unit.
After cooking is done, robotic kitchen puts the container into a lower-level conveyor, and as the container moves away from the robotic kitchen unit, it joins other containers on the merging conveyor.
The merging conveyor takes the form of a belt-driven live roller conveyor, which at the end of it changes into a belt conveyor that elevates all containers back into the inventory.
Task
The task is considered a collection of different types of objects, including order objects and customer demands objects.
Each task object is temporary, and should be linked to a table object during a time when the table is occupied.
An order is an object consistently bound to a table object during the entirety of customers' stay.
The Central System runs a dynamic queue associated with each table object that stores all orders, adding one after placed and removing after customers leave the restaurant.
An order object should specify types and quantities of each dish, individual prices and the total price.
A customer demand is an object permanently bound to a table object. It can take the form of water fills, etc.
The Central System runs a dynamic queue associated with each table object that stores all customer demands, adding one after placed and removing after customers leave the restaurant.
Notes on Implementation
We need to weigh the benefits of outsourcing tasks and taking matters into our own hands. Specifically, we need to understand which companies have created functional models of our desired components and what adaptations we need to make to integrate those devices into our global system. If those existing products are not satisfying, we will instead break a large component into smaller, more manageable pieces and build with our own hands.
Market Research
This short report contains a list of companies with functional models of our desired technical components and encloses individual links.
Central System
Embedded System
The key to successfully implementing this Central System is to interpret it in the context of an embedded system, which takes the form of the Internet of Things (IoT). Specifically, all the robots and robotic kitchens are "Things", and we need to figure out how to build the local network (can either be a local area network or a private cloud, but the cloud option is superior) and provide back-end services.
Private Cloud
Here is a list of top service providers and their comparison in a chart.
Back-end Services
TODO: Find a list of back-end services providers, enclose their links and write a short description of each addressing their solutions for 1) remote server; 2) user access and control; 3) business data analytics.
Robots
Bear Robotics
Penny, a cylindrical robot with a flat top, can take up to the number of all dishes for one table. It knows how to navigate the restaurant floor when obstacles are stationary. Human help is needed in loading and unloading dishes.
CSJBot Amy
Able to deliver cooked meal and collect dishes. Facial expressions observable through human eyes with the help of LED screen, and emotions vary based on interactions with customers. Provided with the touchscreen-based customer engagement software that allows customers to place order and interact with Amy more generally. Uses ultrasonic obstacle sensor to detect and circumvent obstacles. Also uses navigational system to find direction without having to follow any dedicated signs on restaurant floor.
Paaila Technology
Ginger, a waiter robot designed to deliver food across restaurant floor, is powered by swarm intelligence, navigation technology, and digitalized information about the restaurant. It can take a load of food from kitchen to table, knowing where the obstacles are and how to avoid them. Swarm intelligence allows these robots to communicate with each other and with the restaurant Central System, paving the way for task distribution, collision prevention and cooperation.
Robotic Kitchen
Moley
Records the recipe by observing how a human chef would cook it. Stores more than 1000 possible recipes in its inventory. Each Moley cooking robot works as an isolated functional unit consisting of condiments, utensils and ingredients. Makes dishes at relatively high speed. Can be remotely controlled.
Spyce
Ingredients are first prepped in a commissary kitchen before loaded into the robotic kitchen. After customers place their customizations, the machinery is set into motion to mix different ingredients in the induction wok. Customers engage with the robotic kitchen system through a touchscreen-based electronic system. Each kitchen unit is also wired in the way that enables self-cleaning.
Inventory
We will borrow from the idea of a vending machine, which dispenses categorized products based on specific requests. However, our inventory system will differ from a typical vending machine in that it will receive information not from an input keyboard but from the local network. It also doesn't support coin devices. There should be a channel at the bottom of each item storage container that merges all items belonging to the same request into a container waiting at the exit. Items are directly released into the container.
TODO: Find a list of vending machine parts sellers, enclose their links and write a short description for each. Please note that we are not looking for vending machines but vending machine parts, but if you think we can be better off buying manufactured machines and later taking unwanted parts off, you are also welcome to find a list of machine sellers instead.
Conveyor System
Overview
Will have to implement the pallet layer descrambler technology on the sortation conveyor. This technology ensures that a single file forms before the critical sortation decision-point. Other system-wide necessary technologies include brake metering conveyor, etc.
Candidate Suppliers
The following is a list of trusted manufacturers of conveyor system. Making a decision to purchase conveyors is especially important and will demand a close survey of each company.
Customer Engagement Software
mEasyPOS
Offers QR-code based solutions to restaurants. Each QR code can be generated with a built-in location. Customers don't have to download any softwares other than a standard QR code scanner. Waiters are only a button push away.
Waitry
Requires a software to function. Individual restraurants register to become partners so that menus can be uploaded to the service. Supports multiple languages. Allows leaving comments on overall experience and individual waiter performance.
Other uses of QR code in customer relations
Dishwashing and Waste Management
Kitchen Assistant Robot
Developed by Panasonic and IRT Research Institute, this robot is able to identify and pick up any dish from a pile of dishes. After rinsing, the robot puts the dish into the dishwasher. Such robot was developed ten years ago, and by now should have been able to put away all dishes in five minutes after a family of four dine.
Food Waste Management
Machine learning analysis can be a powerful tool for gaining insight into the nature of food waste. By having cameras mounted on robots take pictures of food waste on each table, we can learn some patterns.
In addition to analysis, we can implement many other measures to combat food waste. Refer to this webpage that contains much more useful information.
Tasks before Application
Last updated
