Finding a robot we need?

Today I listened to another person asking the world what sort of robot they should build. They had just told us the features that it would have and want to know what things people might use it for, so that they could find the money to build it. This is a completely back to front approach to robot building.

This is what Caroline Pantofaru, Leila Takayama, Tully Foote and Bianca Soto refer to as ‘technology push’ in their recent paper. An excerpt is quoted below, but I recommend reading the whole thing.

Exploring the Role of Robots in Home OrganizationPantofaru, Caroline., Takayama, Leila., Foote, Tully., and Soto, Bianca Proc. of Human-Robot Interaction (HRI), Boston, MA, p.327-334, (2012)

We present need finding as a methodological tool for quickly learning about a user space, inspiring robotics research within that space, and grounding the resulting research. Much (although certainly not all) of robotics research today is inspired by technology push – a technologist deciding to apply a technology to a problem. User-based research often does not start until after a prototype or system speci fication exists. This is a valuable method as researchers have spent years building intuition for their field.

For robotics research, need finding can provide a complementary, user-driven source of inspiration and guidance, as well as refi ning technology push ideas to better to an application space.

Need fi nding is a method that comes from the product design community [2]. The goal is to identify a set of fundamental user needs of the community a product aims to satisfy. The need finding process is summarized in Figure 2.

Figure 2: An overview of the need nding process

Need finding begins with generating empathy for the user group through interviews, and sharing that empathy with other designers and researchers through conversations and media like videos. This is a concrete and analytic process.

The results from the interviews are then abstracted into frameworks, often presented in graphical form such as the 2×2 charts in Figures 2 and 3. The lessons from the frameworks are then converted to design implications, which are meant to be generative, allowing many interesting solutions to evolve. This process can be iterated and interleaved as necessary. The process is expanded upon below, with a description of our own implementation for this paper.

2.1 Interviews and Observations
Need nding begins with identifying a community of potential product users. A very small team goes out to visit a sample of community members in the places where they do the activities that the product supports. Immersion in the
interviewee’s environment is the key to success and a distinguishing feature of need finding. This immersion inspires the participant to discuss details they might have otherwise forgotten, and allows the interviewer to quickly reconcile the interviewee’s words with reality.

It is important to note that relying on self-reported data alone is dangerous due to people’s poor memories, as well as the social desirability bias (an inclination for respondents to say things they think will
be received favorably.) There is even a standardized scale for measuring a person’s inclination toward the social desirability bias [5]. These problems can be so serious that some usability experts suggest completely ignoring what users say, and instead watching what they do [16]

4 thoughts on “Finding a robot we need?

  1. “Today I listened to another person asking the world what sort of robot they should build. They had just told us the features that it would have and want to know what things people might use it for, so that they could find the money to build it. This is a completely back to front approach to robot building.”

    I assume you’re referring to me here (I’m Tom from that Silicon Beach thread about a robot concept.) I replied in that thread too, but thought I’d follow-up here as well.

    You’re right that this is back-to-front, and that customer discovery should come first. BUT, consumer customers often don’t know what they really want, and even more often don’t understand what is and isn’t possible. This isn’t suggesting they’re stupid, just not up to speed in the field. Robotics is a prime example of this because the field is complex, and public perceptions of it are biased by SciFi and popular culture representations of robots.

    As it happens, I *did* do a lot of customer discovery before putting together the collection of features I described in that thread. Unfortunately, all that effort told me was that people want things that aren’t possible at a reasonable price point. They want a household servant that can cook, clean, walk the dog, fetch beers from the fridge – and they’re not willing to pay more than $1000 for it. Here’s another market survey that suggests similar conclusions:

    If the information I gleaned during my search is true, then we should all give up on consumer robotics for the next decade, because there are no viable customers. What they want cannot be done with today’s technology at that price-point. I am optimistic however, and instead choose to believe that the customers just need to be educated about what’s possible first.

    Next up, I looked at the few things these customers had requested or the problems they wanted solved that *were* viable – Siri-like voice activated assistance, reminders, quick access to visual information anywhere, hands-free computing, games, tele-presence, etc. I connected the dots and realised that none of these tasks require complex manipulators. At about that time, I posted that thread on Silicon Beach asking for even more things that could be done in this space – and that’s where you found me.

    With regards to the paper by Pantofaru et al., I originally took issue with the approach because it jarred with my understanding of needs vs wants. In hindsight this is mostly a syntax issue – replace wants with ‘latent needs’ and everyone’s happy. Overall it’s a solid paper with some interesting conclusions, but in my opinion it’s essentially just advocating using the customer development or lean startup approach, and then targeting this at robotics research.

    If you’re interested, I’ve since refocussed my efforts on an enterprise problem that can also be solved by similar technology. I’d still ultimately like to make consumer robots, but to do so I need to make a long-lasting robotics *business*, and this new focus is a better way to ensure that happens. I’m not exactly ready to pitch for investment at this point, but coincidentally the problem is in exactly the domain that the Citrix Accelerator is focussed on, so I hope to get in touch with Michael shortly as well!

    Thanks for being interested in this stuff. Feel free to email me if you’d like to discuss it further.

  2. Thanks for commenting. As I don’t monitor the blogs 24/7, first time comments are moderated and due to the timezone differences, it must seem to take longer than you may have expected.

    My post wasn’t aimed at you, although your thread on Silicon Beach did fit the description. I’ve just spent 6 weeks hearing pitches and 95% of them have had the exact same issue. “Here’s what I want to build, where are the customers?”

    I’m glad you’ve explained to me the steps you took in researching customer need, but I see a lot more potential in the process than you atm. I also agree with you that need finding is customer validation and lean startup methodology. It’s also behavioral observation, social psychology and anthropology.

    When it’s done well, it works. Most roboticists are not experts in doing this base level human behavioral research, so we should consider getting experts involved. One anthropological tip is, don’t ask questions but observe where bottlenecks and pain points exist. These may be taken for granted. Then think about tools that might fix those things.

    It sounds like you are doing this with your enterprise level solution so I wish you well. I’ve also read the survey you linked to and find it optimistic not pessimistic, but then I know at least one company working on a home/enterprise security bot built out of a roomba/kinect combo, which will have a price point sub $2000, so consumer robotics is well on the way.

  3. I certainly agree most roboticists aren’t experts in behavioral observation, etc. However, many entrepreneurship bigwigs such as Steve Blank suggest that a startup’s founders are the ones that should be doing customer development – even if they’re bad at it initially. Nothing wrong with getting expert assistance, but I think it’s probably important for the founders to learn the lessons in person.

    The tip of observing rather than asking questions certainly rings true. I used to make computer games, and when we brought in people to playtest them, learning to just leave them alone and observe silently taught us so much more than our early attempts. We’d get frustrated and say “no – you’re supposed to pull the lever first” and miss the point that we should have designed things to make that intuitive in the first place.

    I’m very happy to see consumer robots starting to take off. I’ve looked up as many of the companies and ideas from the latest Robot Launch event you helped run, and some look like really neat concepts or products. Not so many in Oz however… yet!

  4. 😛 I absolutely agree that getting founders involved in customer development is integral to the success of the methodology. That doesn’t mean they should go it alone though. The companies, including some in robotics, that find most success using customer development methodology use pair visits (designer/engineer or UX/engineer or ivspecialist/founder etc). It’s a fast track to what you discovered with game design feedback.
    And I’m also looking forward to seeing more of the Aust robot developments later in the year.

Leave a Reply (first comment moderated)

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s