Introduction to Robot Character Design

I believe robots will serve a critical role in society in the future, but perhaps not the roles that we previously hoped for, and for that matter feared. The differentiating factors for adoption of these technologies include yes, advanced technology, but just as important and often overlooked, thoughtful design.

In the next decade I see several applications emerging that bring robots into application domains which add real value to our lives. Teaching children french, and keeping tabs on our pets being just a couple. If we are entering into a golden age of robotics, the history books will be filled with countless one trick ponies, small dogs that can bark twice and then do a backflip. What is around the corner is not only much more impressive, it provides a platform which is capable of creating real value.

To me “real value” is providing new ways to: learn, communicate, play, reflect, and experience the world around us. The last time this happened is when a computer, camera, and a phone became small enough to fit in our pocket. The next time it will happen, is when the small computer grows legs (or wheels), and becomes part of our daily life. Don’t believe me? give it a decade. Ok, lets get started with the most important part of robot character design, the eyes.


Eyes and Eyebrows

If you read the through the last couple decades of HRI (Human Robot Interaction) research you will find a number of papers that focus on the eyes. What happens if the robot makes eye contact with a person? what if they avoid eye contact? what if the eyes are screens? what if they are made of plastic? the list goes on.

Most of my views on robot eyes are from working in a lab that tests robots in HRI studies. I have seen robots with dozens of motors in the face interact with a group of people while forming a strong social bond. I have also seen robots reset, (which looks like they are having a stroke), and see that bond be broken instantly.

Some of the most successful robots take a minimal approach to eyes, C3P0 and R2D2 are prime examples; the Muppets are another. When building social robots, the key is not to paint the entire picture, the key is to paint a few careful brushstrokes and let the person interacting with the robot fill in the blanks in their imagination.

When it comes to more advanced eye and eyebrow capabilities the rewards can be greater, but the pitfalls are also deeper. My view is that it is best to take a minimalist approach, figure out what your robot needs to do and design around that. If you give a robot the ability to squint, blink, and raise its eyebrows, these are all now elements which need to be carefully controlled.


The word affordance is not one that gets passed around lightly in conversation, but it is an important element when designing robot characters. One of the best books on the topic is “The Design of Everyday Things” by Don Norman.

Norman’s thesis is that the ability of an object to perform certain tasks influences how we interact with it. A door that has a push only lever on it, or a tea pot that requires two hands to use are a couple examples.

Robots also have affordances, and on a deep subconscious level form our views of what the robot should be able to do and how we should be able to interact with it. Below are a couple robots with a single purpose, one is named DJ and the other Optica.

They both have affordances, and by nature of their form suggest a number of things. One has the ability to play music, and one likes to take pictures. These are simple examples, but hopefully get the point across.



When assigning human traits to non-human objects, the word anthropomorphism is a bit of a catch-all. There’s no shortage of robots that attempt to directly mimic humans in form and behavior. It is worth mentioning that this is a tricky thing to get right, and the path is filled with pitfalls.

If you are designing a humanoid robot, my advice is to shoot for non-threatening. Honda for instance made a humanoid with the height, and head to body ratio of an adolescent child. Many of the most successful humanoid robots use a number of tricks like this to create a welcoming appearance.

Although some may disagree, my view is that it is best to avoid the uncanny valley at all costs, which is the place where “this is real”, and “this is fake” collide to create disgust, distrust, and dissonance. Think “Lars and the Real Girl.”

Evolutionary psychologist have a number of theories as to why the uncanny valley even exists, and most of them relate back to needing a high sensitivity to detecting people who are “maybe dead”, or “maybe have a weird skin issue”, (just a couple examples.)


Physical Limitations

When it comes to affordances, it is all about doing the most with the least. If the form of a robot suggest a high level of capability, but then underdelivers, a low bar is set. On the flip side, a robot that can impress with very little will often be held in high regard.

One of the tricks robot designers use to generate an emotional response is introducing a physical limitation. A simple example to play with would be a robot with arms. Imagine two robots, one with big gorilla arms and one with tiny toothpick arms.

Each creates a different emotional response. The gorilla armed robot may be just as good at bartending, but so what? he has big strong arms. Now look at little toothpick arms go to town on those Manhattans! Amazing, the reason why? physical limitations.

Another example would be a small statue of Micheal Jackson that does nothing, and a coke can that can dance like Micheal Jackson. One is amusing, one is absolutely mind-blowing. Lesson: make limitations an opportunity to deliver on expectations.


Animal Mimicry

When it comes to animal mimicry, we are dealing with very powerful emotional responses which are often hard-wired into the human psyche. An example would be a robot snake or spider, which sends a very clear “not to be trusted” signal.

As an extension of physical limitations, animal mimicry adds a new twist. A dog robot may get some “adorable points” by association, but if it can barely walk, the robot is not that impressive. A turtle robot that can barely walk on the other hand may be adorable and impressive.

One interesting example of animal mimicry done well is “Paro” a robotic seal that was created mainly as a companion for the elderly. The robot proved successful in being a cute fluffy friend capable of boosting morale in elderly homes, but also had an unexpected flip side.

If you read through the papers on Paro, you will find numerous accounts of elderly users giving Paro the seal a bath, which would be fine if Paro were a seal, but not great for a robot. Animal mimicry is a powerful tool but it may be best to use this design influence like salt, just enough, and no more.


Taking a Cue from 2D Animators

If you decide to get into robot character design, the best thing you can do is grab a copy of “The Illusion of Life: Disney Animation”, which is a book written by Ollie Johnston and Frank Thomas, two prolific animators from Disney.

Many of the principles from this book apply directly to robot character design, and can help build a framework for building robots which also have the “illusion of life”. Below is an illustration of one the main principles in action known as “squash and stretch”.

When animating a rubber ball, the basic technique is that the ball has form, and the form changes with action. When bouncing, the ball squashes down, and stretches out. There is nothing that says a robot needs to do the same, but when following the same rules, a robot can also come to life in the same way.


Final Thoughts and Exploring Form Through Storyboarding

In the end there are several questions that need to be asked before designing a robot character. The primary being: what is it’s purpose? who is the intended user? and how much should the robot cost? After that, it is a creative process to find a robot that can fit the role.

One of the best ways to go about such an endeavor is to test a robot’s emotional range through storyboarding. Come up with an adventure that puts a robot though a number of interactions. Perhaps your robot meets a new friend, takes a nap, and then find a new toy.

Below is an example of a sketching session for such a robot adventure. This is the perfect exercise to give a design team, and the perfect thing to pass to an engineering team to see if such a design is possible. Thanks for reading and happy robot building.