Week 11+12: Toy Redesign and Test

This week we redesigned our toy based on the user testing feedback, looked into research on similar toy concepts and tested the new interaction during take your child to work day.


One of the biggest takeaways from the user testing was that the toy needed to more explicitly encourage kids to explore the outdoors. One suggested we received several times was to create a scavenger hunt. We decided to use machine learning to recognize objects and guide the child on a scavenger hunt. Based on this updated concept, we created a new interaction model:


We made an updated moodboard, sketched a new design for the toy and based on that made an initial 3D model. We decided to update the material from felt to silicone because the and we received feedback that the felt would probably got too dirty and would be hard to clean.

Artboard 1 copy@2x.png


In order to ground our project within the smart toy field in terms of both concept and function, we looked into research that has been done on developing outdoor smart toys. We read the following papers:

Scratch Nodes: Coding Outdoor Play Experiences to enhance Social-Physical Interaction (Hitron et al. 2017)

A Little Bit of Coding Goes a Long Way: Effects of Coding on Outdoor Play (Hitron et al. 2018)

Evaluating Outdoor Play for Children: Virtual vs. Tangible Game Objects in Pervasive Games (Soute et al. 2009)

Heads Up Games: The Games of the Future Will Look More Like Games of the Past (Soute and Markopoulos 2007)

Heads Up Games: combining the best of both worlds by merging traditional and digital play (Soute et al. 2009)

And some reports on kids and nature:

Do our Kids Have Nature-Deficit Disorder?

Last Child in the Woods

Outdoor Alliance for kids

Green Teacher

Learning with nature book

Scholastic Outdoor Alliance

As we continue this proejct, we would like to write a paper about our toy using similar methods to the ones in these papers. Our research plan would include the following elements. This is something we have decided to discuss further once the semester ends and we have our final toy.

  • Lead outdoor play session

  • Outdoor engagement and physical activity observation

  • Survey before and after about feelings towards playing outside

New Prototype


For our new prototype, we were relying on the phone to run the machine learning program. We are using p5 in order to allow an interaction with a button the screen, so we first we tried using the p5 generated voice for the prompt, but it sounded too robotic. Instead, we decided to switch to using voice recordings corresponding script and recorded voices for. Thank you to our classmate Lillian Ritchie for being the voice of our toy!


Using p5 and Google’s teachable machine, we trained the toy to recognize five objects:


User Testing

This week Tisch hosted a take your child to work day, so we demo’d this interaction with the kids there. We set up the computer with an external camera and placed the trained objects around it. We showed the kids and . Most of the kids were five or younger and enjoyed searching for objects and seeing if the computer could recognize them. We also spoke with parents and received positive feedback from them on the concept. Many like that the toy would get their child outside and off screens and also liked that it was toy that they and their child could play with together.



For the demo in class, we fabricated a low-fi toy prototype.

In our demo the toy identified one object, but had some bugs as the interaction went on.

Our next steps are 1) to update the code and stress test it and 2) fabricate the toy out of silicone

Week 10: State of AI Research

For this week’s assignment, I read a paper titled “Envisioning AI for K-12: What should every child know about AI?” The goal of this paper was to be a call to action and put forth an optimistic view of what AI education should look like in our near future. The motivating statement was shocking to me because I had never conceived of the future of AI in quite this way:

“For many in this generation, AI will be an often overlooked, magical force that powers their lives much as electricity, the internal combustion engine, and networking technology power ours.”

The authors first describe the state of AI education for kids and then outline five big ideas in AI and what students should be able to understand an do related to each one in the four grade bands (K-2, 3-5, 6-8, 9-12):

  1. Computers perceive the world using sensors

  2. Agents maintain models/representations of the world and use them for reasoning

  3. Computers can learn from data

  4. Making agents interact comfortably with humans is a substantial challenge for AI developers

  5. AI applications can impact society in both positive and negative ways

I believe that the fifth big idea is the most important one because without it, their “blue sky” predictions about AI will not be realized. Bias is mentioned only once in the paper and I think that big questions such as this and also about ethicsn and human rights need to be front and center in this work. Language in these curricula will be very important as well in creating new standards and conventions and will even extend to how students conceive of themselves, other humans, and the natural world. I would encourage the creation of these curricula to be an interdisciplinary endeavor (including other disciplines such as sociology and philosophy) rather than having it developed solely by computer scientists. Overall, I find the work that the authors of the paper are doing encouraging!

Week 8: User Testing

This week we went to a school and tested our toy with two groups of 3rd and 4th graders. Each group had 4-5 kids in it and most were boys. We had about 10-15 minutes with each group.

Drawing activity and use of the toy

Screen Shot 2019-04-11 at 5.52.55 PM.png

In order for the team to start the conversation with the children, we decided to create an activity that would engage them in learning about the toy. We created a coloring book style drawing of the toy and used this activity for two main reasons:

1. Collect feedback on the names that children like for toys like ours

2. Motivate them to color the drawing using as many colors as they wanted, in order for them to later use those colors to be scanned by the toy and generate sound with them.  

It was important for us that the kids trusted us and we focused on empowering them so that they could play with the toy in a very loose and honest way. The main purpose of the exercise and its respective observation was to be able to introduce the children to the use of the toy in a progressive and more involved way.

Key takeaways from the activity:

  1. Kids engaged very fast with the coloring activity and didn’t need much instructions to do so.

  2. They used the colors of the draw to scan them with the toy however they didn’t immediately understand what was going on (the colors changing the notes being emitted by the toy).

  3. Some children were methodical in figuring out the toy, scanning one colored crayon at the time, while others were more exploratory and scanned book pages with many colors at once

  4. The kids were more motivated to scan other colorful objects such as the books and the balloons that were also settled in the table where we performed the test.

  5. A kid with learning disabilities was the one who engaged in a better way with the toy, as researchers we have the theory that this might have been because our target audience was younger kids and the kid with learning disabilities might have had a similar interaction than our originally intended target audience.

In terms of the design, we observed that the mapping of colors to ones which have a strong symbolic association (red, green, yellow) creates the connotations of ‘right, wrong’ in the minds of children. We also observed that the sounds emitted need more variety for both keeping children’s interest and having a clearer correlation with the colors they are scanning.

Direct questions to children

During the playtime with the toy, we asked each group of children the same set of questions in order to understand how they conceived of the toy and interaction. Below is a summary of the questions and their responses.

The children quickly learned the interaction and several wanted to know how it works. We did have to do some prompting to show the children additional ways to explore with it, so one take away was that it may help to design into the toy interaction some prompts for exploration and have more narrative or framing around that.

How would you describe this toy to your best friend?

  • Music player

  • Smells stuff

  • Noisy

  • Lights up

  • Colors make the noise

  • I’ll teach them how to use it

  • This is a recorder

  • It makes sound

  • It’s fun because it can make music with color

Would you want to play with this toy again?

  • Yes

  • I usually play with toys to make fun of it

Where would you want to play with this? Inside? Outside?

  • I want to take him home. I want to sleep with him.

  • Put it on some piece of paper, color the paper and put the nose on it/ draw on the wall

How does the toy make you feel?

  • Good

  • Oof

  • Confused

  • Want to see what will happen

  • Interested

Other Quotes

“It thinks bad of you because it’s eye is read”

“He’s a better singer than you’ll be


“I want to use it to annoy my mom”

“Can I take him home?”

“Try green!”

“I found something he doesn’t react to!”

Group Reflection

After the playtesting, we came back together as a group and reflected on each toy and talked about AI and smart toys in general. It was enlightening and a bit disheartening to hear about the children’s current play practices, especially the heavy reliance on youtube and ipads. Below are some notes from the discussion:

Notes on our toy

  • It should have 2 noses

  • It should look more like an elephant

What is AI?

  • A computer

  • Google is controllable by another robot and another robot controlling that robot

More kids will play with tablets and phones

  • You can play with multiple games

  • Mostly I watch youtube (slime videos, diys)

  • More advanced than toys

What are your favorite toys?

  • Hamster

  • Chess

  • Tablet → phone → nintendo

  • Books

  • Stuffed animals

  • Jumping

What is most challenging for you?

  • Swimming

  • Listening to teachers

  • Waking up

  • Annoying mom

  • Getting bad grades

  • Going to school

Week 7: zipzop

Soundy Presentation.jpg

Based on the feedback we received during the user testing sessions, we updated our smart toy in the following ways:

  • Created user personas

  • Redesigned the form

  • Added color to sound sensing

  • Created a narrative around the interaction

You can see the presentation outlining these updates in more person here. Below we have presented the new concept for our toy.

User Personas

We created the below persona for our child user and their parent. We want this toy to be as accessible as possible for people with and without tech backgrounds. Because one of the goals of this toy is to help and encourage kids - and especially kids who are “nature illiterate” - to play outside, we used New York City as our example location.

Soundy Presentation (1) copy.jpg
Soundy Presentation (1).jpg


Based on feedback to make a narrative around the sound and color collection, we modified the scavenger hunt idea to create a companion that the child can build a story with. The toy prompts the child to explore their environment and collect colors for parts of the story:

Soundy Presentation (2).jpg
Soundy Presentation (3).jpg
Soundy Presentation (4).jpg
Soundy Presentation (5).jpg

Week 5: Research Paper

For my research paper I selected a paper from the 2011 Interaction Design Foundation Conference titled “Steps Toward Child-Designed Interactive Stuffed Toys.” I selected this paper because the abstract sounded relevant to the toy prototype that Veronica, Arnab and I are working on. While many other papers centered around some common topics (tangible programming, math concepts, and augmented reality), it seemed to be the only paper that dealt with soft or plush toys and I was excited to see how they were thinking about this material.


This paper outlines a toy “system” called Plushbot developed to allow children to design and build their own plush smart toys using the recent developments in microprocessors and sensors for soft or textile wearables. Children’s toys - especially those that the authors call “character toys” -  have taken advantages of these small new devices for physical computing and the authors want to give children the agency and support to use their creativity to make their own.

Among others, Furby and the Aibo dog are referenced as precursors to these kinds of toys. More recent toys such as Sniff (a dog that we actually used as a reference for our prototype!) that uses an RFID sensor and Probo (a huggable elephant) are referenced as new developments in toys with “sensory abilities.” The authors argue that our current moment in history is a unique because the hardware is cheap and accessible enough for the first time to allow kids to design and create their own toys. They propose their Plushbot software as a way for kids to do this.

Plushbot uses Arduino and LilyPad boards and is designed to work with other sensors and electronics made for wearables. The authors point out that the community of tinkerers focusing on soft toys is small (Plushie is one example they cite) and hope to expand this with their software. Plushbot consists of two main web interfaces: the pattern interface and the playground interface.

Pattern interface

Pattern interface

Plushbot’s pattern interface allows children to draw out a fabric pattern for their toy.

Playground interface

Playground interface

The playground interface allows the child to add microcontroller, circuitry and sensor elements to their pattern. This software is also designed to integrate with laser cutters.

Screenshot 2019-03-07 19.39.40.png

The final phase is the construction phase (it’s not clear from the paper if this is done by the child or how this is supported by their system).


This software looks like an amazing planning tool, but seems pretty advanced for a child. Even though there is a high level of support and assistance from the software, it’s a lot to learn: design, circuits, fabrication. The authors say that they plan to test this software and process with a group of middle school children, which seems like the right kind of supported environment for this kind of learning. Some of the steps seem pretty challenging - for example, there is a note about sewing extra patches onto the conductive thread to prevent short-circuits. In addition, I think it can also be challenging to design a 3D object in 2D and understand and see ahead of time how those two perspectives map together - it’s a type of thinking that has to be developed and could be challenging for first time users.

It seems that there’s a bit of a mismatch with the software design and intended audience - I think that middle schoolers are the right age group to be able to understand the concepts involved and execute them, but I’m skeptical that they are an age group that would want to design plush toys for themselves. In terms of future work, the authors also describe making their own development environment for this system instead of using Arduino. I think that this could be one way to break down the complicated process of making a toy and make it more accessible to more people.

Since this paper is from 8 years ago, I thought it might be out as a consumer or educator product. When I tried looking up the toy, I only found a beta website. However, I was able to login and play with the software! It seemed to have some glitches, but it was fun to see how they were thinking about designing a system to make this process available to young tinkerers.

The landing page

The landing page

I started to make an owl!

I started to make an owl!

Week 4: Low-fi Toy Prototype

Our assignment this week was to create a low-fi toy prototype for a specific user in a specific environment.

User Persona + Environment

After discussing our interests, we centered on the idea that we wanted to make a toy to help a child (up to about age 10) explore the outdoors.

We started by brainstorming and writing down the core concepts we wanted to include in the toy design:

Some initial responsive forms we brainstormed

Some initial responsive forms we brainstormed

As we talked about our experiences exploring the outdoors and looked at references, we started to convene on the idea of a scavenger hunt. More recently, we all made sound walks last semester in our Intro to Video and Sound classes add found that it was a fun and exciting way to explore a space using a different sense - and that we all felt we gained a new perspective from it. We liked the idea of creating a toy companion that would help a child do this.

To create this experience, the toy would be designed to take in one type of sensory input and provide a different sensory output. We discussed the following as possible sense mapping pathways:

  1. Sound to color

  2. Color to sound

  3. Texture to pattern

For this prototype we decided to focus one one: sound to color.

We also outlined a couple goals that we wanted to accomplish in order to guide our design process:

  1. Input → output: are kids interested in exploring the world this way of exploring the world?

  2. Form → Hold? Wear? Augment? What type of form factor is most appealing to kids for this type of companion?

After talking with our professor Stefania, we decided that using a phone to do the computation would be best for a prototype so we could iterate quickly.

We started with creating a mood board and some initial sketches of possible designs for a toy companion that a phone could fit into:



Once we decided on a design, we got some felt and velcro and set out sewing and started coding a sketch in p5js that would convert sound to color!

We tested sounds and songs with the visualization:

The stomach lights up in response to sound

The stomach lights up in response to sound

The hands attach with velco so that the companion can be worn around the body or hugged.

The hands attach with velco so that the companion can be worn around the body or hugged.

We’re excited to test it out!

Week 3: Toy Research

This week our assignment was to research a toy and see what the company, parents and kids say about it.

Animated GIF-downsized_large.gif

I went to the the NY Toy Fair (the largest in North America!) with some ITP classmates last weekend and was overwhelmed by the quantity and variety of toys. While we were there, we were on the hunt for smart toys. This was a bit difficult because there wasn’t a devoted smart toy section, but we did happen upon some pretty cool (and some really well designed!) toys.

One in particular that I liked was a toy called Cubelets that allows kids to build robots from modular parts. All of the parts - action cubelets, sense cubelets and think cubelets  - are in the shape of cubes and they connect with magnets. They remind me of Little Bits because of the physical and modular nature of them and also represent a trend of embodied or tangible computing that I and I think many of us at ITP believe is important. The cubes can also be programmed using Blocky or C. I didn’t get to play with them too much at the Toy Fair, but from what I could see, they seem very intuitive and allow kids to make things quickly and use their imagination. So let’s see what other people think about them!

What the creators say

The website looks very modern and sleek, which matches the minimalist and shiny design of the cubes. Their tagline is “Robot Blocks - The building blocks of better thinkers.” The featured “About Cubelets” video however starts a man who lives in a cabin on a mountain who says he is the inventor and compares cubelets to ecosystems in nature, which doesn’t seem consistent with the rest of the messaging, although I like the comparison!

Screenshot 2019-02-22 00.39.21.png
Screenshot 2019-02-21 23.45.51.png

The website says that the toys are designed for kids 4 and up. They sell different sets and kits with combinations of cubes ranging from $130 to $2,100. They also sell packs specifically for educators with groups of students ranging from $1,390 for 6 students to $3,900 for 12 students. They also have 50 free lesson plans for teachers with students ranging from pre-K to high school and workshops. It also looks like there is an online community for educators - many people have posted videos on the website of how they have used cubelets in their classrooms. The website also has a very active blog with examples of how people are using cubelets.

Screenshot 2019-02-22 00.39.43.png

The website describes the blocks as:

“Designed for all learners”

“Complex lessons made simple”

“Knowledge made visible” -

“Creating connections”

“The building blocks of programming”

The language is definitely geared towards learning STEM concepts, trying new things using both creativity and critical thinking. This language seems targeted towards parents, but maybe even more so for educators to help their students learn - they say they also provide “progress monitoring trackers” to help teachers know if their students are understanding concepts. The website highlights communication and teamwork among students and easy integration with existing curricula. So although this is marketed as a toy, it is equally or even more heavily marketed as a learning/teaching tool.

What parents + kids say

I didn’t find many reviews and I didn’t find any explicitly from kids, but here are some main take-aways from what I found:


Theimums said that she likes that the intuitive and modular design because it’s easy for young kids with small hands (her boys are in preschool). Because the blocks can only be connected one way, this is a simple “cause and effect” that helps young kids build.

Screenshot 2019-02-22 00.46.09.png

One STEM teacher said in a YouTube video that she really liked that the cubes are durable because her students drop them all the time and they still work!

A dad on RobotShop said that he thought that the parallel programming method of cubelets was confusing for adults who are used to developing robots by building first and then programming, but that this came naturally to his kids!


Theimums said that the app experience was great for her as a parent with the options it gave for reprogramming the robots without taking them apart. She also liked that the toy teachers her kids STEM principles. The dad from RobotShop also liked this aspect and echoed the “cause and effect” learning method by saying that it taught his kids to make hypotheses and test them.

Screenshot 2019-02-21 23.58.04.png


theimums said that her kids didn’t want to stop playing when she said that their 25 minutes were up. The dad on RobotShop also said that his kids - a 3 year old boy and 6 year old girl - keep returning to play with them, although the boy tended to be more interested. He said that the robots enable different kinds of play - sometimes they play “intensely” and other times more “casually.”

What I thought most interesting was that he said that he felt there was a strong emotional component to the robots - that they created joyful discovery for his kids.


On Amazon (here and here), a few reviews mentioned the high price tag. However some people thought it was worth it - one reviewer who was able to buy more parts gave a great review and a former educator gave a 5 star review for their ability to teach STEM concepts:

Screenshot 2019-02-22 00.10.08.png
Screenshot 2019-02-22 00.12.08.png

Week 2: Learning with Cozmo

Such a cutie!

Such a cutie!

This week our assignment was to work in small groups to design a learning activity with one of the smart toys we tested and played with in class. Our group chose Cozmo.

The Persona

We set out to make a game for a 7-9 year old child who likes competitive games and using their imagination to create new worlds. This child is also a tactile and hands-on learner who likes to use their creativity and ingenuity to build, tinker and create projects.

game rubric.png

The Design

We wanted this game to be:

  • Active: We wanted kids to move their bodies when playing this game to encourage them to think about how they and their toys interact with their environment, instead of being isolated from it.

  • Collective: The games we played with Cozmo in class were more targeted towards an individual interacting with Cozmo, so we wanted to find a way to have kids play with this toy together.

  • Ruled: We wanted to make a game that had ground rules but still allowed for creativity inside those constraints.

  • Physical + Mental: We wanted kids to move and explore their own space, something that there is a danger of losing with smart and virtual games. We also wanted kids to learn to code in an exciting and novel way.

  • Gender Neutral: We believe these toys and game should be for everyone.

Our goal was to make a physical game to help young hands-on learners begin to code. Scratch is an incredible learning tool for kids, but most of the projects are online and we wanted to create a learning experience to teach them how the physical world can be designed, changed and interacted with using code. In addition, based on our playtest with Cozmo in class, we also wanted to challenge ourselves to create an interactive learning game without depending on the Cozmo smartphone app, which we felt interrupted the experience of engaging with Cozmo.

The Game

1. Design the course

1. Design the course

2. Program your moves

2. Program your moves

3. Drive!

3. Drive!

This is a 2+ player obstacle course race - designed by the kids’ imagination and built with household objects. After creating the course, the kids are instructed to place Cozmo’s cubes along the course - the goal is to get to the finish line first, but tapping these during the race will get you extra points. A player could theoretically win even if they finished last, but tapped more cubes. Once a cube is tapped by one Cozmo, it cannot be tapped again.

The kids place their Cosmos on the starting line and study the course. The challenge is to find the shortest/quickest path to the finish line and get the most points by weighing the option of detouring to tap cubes. Using physical codeblocks (inspired by LittleBits and Scratch and yet to be developed), each child programs a series of commands for their Cozmo to navigate through the course. We came up with the idea of physical code blocks because we wanted to find a tangible way for kids to create code offline. Looking at the design of the Scratch, we imagined these blocks of code could be fabricated to be modular “puzzle” pieces similar to LittleBits or Legos.

This game is designed to help children to learn:

  • Coding skills

  • Spatial awareness

  • Logic

  • Exercising their imagination

  • Collaboration