A smart and healthy cat-feeding machine

Case study of a project with

Project genesis

Pierre, Yoan, Delphine and Julien contacted us to build a solution to improve and simplify the feeding process for cats. Yomy's idea was to build a connected dispenser that distributes both wet food and kibbles, a wide range of wet food for cats, and an app to monitor and analyze the feeding behavior of the cat. Yomy's approach was to involve smart features in their product design, such as a bowl equipped with a weighing system, a food drawer that optimizes food storage and minimizes odors, and a camera with night vision and a wide angle to detect when cats are hungry.

Our mission was to spearhead the R&D efforts, design and develop a fully integrated prototype of the Yomy machine and app, and ultimately manufacture a first batch for beta testing.

A smart and healthy cat-feeding machine

The challenges we faced

In order to help Yomy team launch such a venture, we had to imagine a solution able to efficiently distribute both wet food and kibbles. Many usage, technical or regulatory constraints had to be taken into account. How can we build such a machine rapidly?

  1. R&D on wet food dispensing: Designing a solution for wet food dispensing system is a complex challenge. Unlike dry kibble, wet food is stored in pouches or containers, it is a softer and more malleable substance that requires a different approach to dispensing. How do we imagine a solution that is reliable, hygienic, easy to manufacture and that generates less than 5% waste?
  2. Hardware integration: Building a dispenser that can accurately dispense both wet food and kibbles requires careful consideration of mechanical design and actuators/sensors integration, especially when it comes to the food distribution system. How can we imagine a reliable system that can durable withstand daily use?
  3. Cat detection: Cats can be hard to detect in a home environment. Capturing clear and accurate images anywhere, anytime and developing and training machine learning algorithms to accurately recognize and track the cat is a big challenge. How do we build a reliable solution to ensure (hungry) cats will always be detected?
  4. Safety: How do ensure that the machine's moving parts, such as the food dispenser and drawer cannot injure a cat or a baby if they were to accidentally get too close to the machine while it is in operation?
  5. Ergonomic and cleanliness: The machine needs to be designed with some easy-to-remove and easy-to-clean components. Ensuring that users are able to clean the machine safely without damaging any of the internal components can be tough. How do design a system optimized for cleanliness?
  6. Connectivity: A crucial aspect of Yomy is to communicate with a Cloud-based server and cat parents through a mobile app. The machine needs to connect to Wi-Fi and transmit data such as photos and telemetry in real-time. How do we optimize the IoT connectivity of the machine?

The solution we built

An elegant, ergonomic and reliable machine

  • Capable of dispensing both kibbles and wet food for cats
  • With 2 modes of operation: scheduled (at fixed slots) and free-feeding (when the cat is hungry)
  • With a computer vision algorithm that detects the presence of a cat in front of the machine

An IoT infrastructure to process and store data

A mobile app that allows cat owners to configure food distribution, scheduling, and other settings

A back-office for Yomy team to manage their fleet of machines and manage their beta testing

Over the course of about a year, we helped Yomy team go from a concept to beta testing and Seed funding.

Weeks 1 to 5: Getting the concepts right

Milestone 0 aKa Find your Spark

We needed to develop a V1 prototype with main functions for dispensing wet food:

  • How can we empty a pouch?
  • How can we build a system to manage 2 pouches ?
  • How can ensure proper sealing of a pouch between 2 distribution sessions?

We explored several options and built simple prototypes to validate feasibility. We opted for industry-standard gusset pockets, which are pyramid-shaped with two slopes. After thorough testing, we were able to successfully build a prototype that met our requirements.

💼: Validate feasibility for wet food distribution with 2 pouches

🔧: First prototype / test bench

💸: 50K USD invested to date

Weeks 6 to 10: Internal testing

Milestone 1

We conducted internal testing and optimization on the V1 prototype:

  • Distribution reliability / loss
  • Eease of use and safety
  • Optimal sachet size, taking into account industrial constraints for filling and sterilization

After conducting these tests, we were able to identify and address any issues with the prototype's functionality and design. We also gained valuable insights into the most effective sachet size for the product. Overall, these tests allowed us to refine and improve the prototype before moving on to external testing."

💼: Iterate on R&D and system reliability before development of the first integrated prototype

🔧: Improved prototype and technical choices

💸: 150K USD invested to date

Weeks 11 to 15: testing, testing, testing

Milestone 2

We were not satisfied with quantity of wet food loss with the actual system. So we built several test benches to evaluate alternative technical solutions. We also designed the optimum profile for the Yomy sachet, including the width, height, funnel, and neck. These test benches allowed us to thoroughly evaluate and compare the different options, and ultimately choose the most effective solutions. The resulting Yomy bag design and dispensing process were more efficient and user-friendly than the previous prototypes, making it an ideal solution for dispensing wet food.

💼: Iterate on R&D and system reliability before development of the first integrated prototype

🔧: Improved prototype and technical choices + overall cat feeding experience

💸: 250K USD invested to date

Weeks 16 to 20: test on site

Milestone 3

After testing our prototype in our lab, we decided it was time to move on to real world tests. We built a second iteration of the prototype which added two features:

  • Scheduled meals
  • Ability to call the cat

💼: Iterate on R&D and system reliability before development of the first integrated prototype

🔧: Figuring out how to detect the cat, we explore Bluetooth Low Energy (BLE), motion detection, and AI-based techniques for identification.

💸: 90K USD invested to date (30K USD for the milestone 3)

Our first tester and the prototype in the background

Weeks 21 to 25: Launching the alpha testing

Milestone 4

We perform test benches for cat recognition through AI (based on an upgrade of the V2 prototype), and for the integration of the drawer.

💼: The core of the product is developed, we continue our R&D to add features, improve system reliability, and validate our main features

🔧: We decided to stick with an AI system that detects the cat through an integrated camera.

💸: 120K USD invested to date 30(K USD for the milestone 4)

Weeks 26 to 30: Integrating new functions and new design

Milestone 5

We build a third functional version of the prototype that incorporates four additional features:

  • Drawer
  • Control via mobile app
  • Cat recognition through AI
  • Dry food dispenser

💼: The core of the product is developed, we continue our R&D to add features, improve system reliability, and validate our main features

🔧: We test the four features we added, this leads us to make several changes in our design: improving balance, camera angles, camera lighting, and closing of the drawer.

💸: 135K USD invested to date (15K USD for the milestone 5)

Weeks 31 to 35: Improve and duplicate

Milestone 6

Alpha testing leads to optimisations of our third version prototype, developing it further, and build five dispensers for alpha testing.

💼: We start deploying the product to a very small amount of users to gather feedback and more real world experiences.

🔧: We improve the way we measure the amount of food dispensed through a balance in the drawer, update the regression coefficient, and improve the insulation of the kibble tray.

💸: 150K USD invested to date (15K USD for the milestone 6)

Weeks 36 to 40: Road to beta-test

Milestone 7

Based on the feedback from the alpha tests, we worked on improving the reliability of the firmware and changing the design to produce 10 prototypes to conduct beta testing.

💼: Our first batch of testers gave us valuable feedback which we've used to improve the design, we distribute the product to a larger group of beta testers.

🔧: We improve the AI detection, introduce free-feeding, upgrade the firmware, improve the mobile app, add scripts for admins, and debugging tools.

💸: 170K USD invested to date (20K USD for the milestone 7)

Lot's of cats were involved in the beta testing phase of the project

Weeks 41 to 45: Production

Milestone 8

We build 10 devices of the beta test version to gather feedbacks, improving the software as well

💼: We manufacture our first batch of 10 units for our group of testers.

🔧: We improve the software and start production of 10 units.

💸: 190K USD invested to date (20K USD for the milestone 8)

Weeks 46 to 50: testing

Milestone 9

At this point our testers give us feedback on their experience using the feeder and we have a good idea of the important features, the ones that need tweaking and improvements.

💼: We manufacture our first batch of 10 units for our group of testers.

🔧: We work on improving the AI and doing firmware improvements.

💸: 200K USD invested to date (10K USD for the milestone 9)

Weeks 51 to 55: Preparing the indus

Milestone 10

All our tests are done, we have a very good idea of what the key features are, we spend this last milestone getting the product ready for large scale industrialisation. This means getting the user experience right and making an elegant design to fit the product in a house.

💼: All tests are done and the product is ready for large scale industrialisation

🔧: Focus is on the user experience, parts are reengineered based on feedback from testers, we interact with designers to get the finishing touches right.

💸: 240K USD invested to date (40K USD for the milestone 10)

Here's what the final product looks like

Project Journey

The team from Yomy came to us with a super simplified Proof of concept and starting from here, we had to go to a machine ready for beta-testing and pre-industrialization. For almost a year, Sparkmate led the development working closely with the Yomy team and testers. Here is a breakdown of our collaboration!

Weeks 1 to 5: Getting the concept

Let's make it happen

Spill the beans about your problem, challenge, intuition... and we'll bring a solution to life at lightning speed.