Introduction
Most of the social robots are going through crowdfunding (Indiegogo being the main platform).
However it is important to notice that ALL of them are delaying the shipping to their backers by a significant amount of time (6 to 12 months or even years in the some cases).
The consequences are not negligible:
- Lot of backers ask for refunding meaning that the real number of people willing to acquire such social robot is diminishing;
- Lack of trust/credibility in the ability to deliver such social robots and new projects will not be able to reach their funding target through crowdfunding like Yumi from Omate which failed to raise US$ 100 000 despite quite attractive features and a nice design.
It is important to understand why because if these companies are not able to deliver, the “social robot” market will be badly considered by the public and it will be harder for new startups to come with a real product and convince that they can deliver in time.
In order to go further, let us pick the most popular projects that have raised more than US$ 500K through crowdfunding the last 3 years:
Except UBTECH which is an established company, the three other ones are start-ups but the story behind them is totally different. Jibo Inc. shortly after an amazing crowdfunding campaign, was able to raise US$ Millions while Blue Frog Robotics and InGen Dynamics are struggling to deliver their robots and find investors.
We will see that key factors are background of the founders, objectives of the company and the limit of crowdfunding platforms.
Social robot functionalities and company Objectives
In the table below, I highlighted the features of the social robots we are focusing on.
In terms of features, they are more or less equivalent in terms of “high level functionalities” such as personal assistant, get information from Internet, home appliance, home security, entertainment… but they are not really clear about the details, for instance which API are supported regarding home appliance? Of course anything is possible because it is just matter of programming and they will rely on the community of backers to expand the number of services the robot will offer.
However the number of proposed features is really huge, you can almost do anything. But in term of software development, i.e. developments, integration and tests it is a lot of work, especially if you have very limited resources.
The differences reside mainly in the mobility, BUDDY is equipped with wheels, AIDO is a ballbot (clearly a differentiator factor with respect to its competitors but at the price of a more complex robot to control), ALPHA2 is equipped with legs. Only JIBO cannot move but it is provided with an articulated body that will be used to exhibit “emotions”.
Add mobility to a social robot is not an easy choice. If you have children, especially young ones, you know that when they play, they put almost everything on the floor and thus for a robot equipped with wheels, a ball or legs, moving is a very very challenging task!
In the same spirit if you want the robot to move in your home in order to secure it and detect intruders, you need:
- To provide a map of your apartment/house in a way or another. Automatic mapping of your apartment/house by the robot is out of reach regarding the price of such robot.
- To localize the robot with respect to its environment (odometry for BUDDY and may be Kalman filtering for AIDO) – with the drawbacks inherent to these methods leading to loss of precision with time.
Ingen Dynamics or Blue Frog Robotics are not particularly talkative on this subject.
This is why having a robot that you put on table near you is not a so bad idea because you get rid of all the problems related to mobility and the frustrations of the users that will realize that the robot will be able to wander in a clear space in most cases.
JIBO the most delayed project, the most funded, the most ambitious
It is quite difficult to classify JIBO. JIBO Inc. the company which is making JIBO has been founded by Cynthia Breazeal a leading researcher in Social robotics and the Director of the MIT Media Lab’s Personal Robots Group. Cynthia Breazeal has spoken at a number of prominent global events including the World Science Festival, the World Economic Forum, TEDWomen and many others. She is familiar with and appreciated by the Media community.
When JIBO made its apparition on Indiegogo in 2014, it was not a surprise that all the “tech&Geek” magazines talked about it. As a consequence the campaign was a big hit and we have to acknowledge that JIBO is the igniter of the Social robot wave we are witnessing today.
However two years have passed and on one hand JIBO is in development stage still and on the other hand international bakers received an email on August 2016 telling them that JIBO will not be delivered outside of USA/Canada. The invoked reason was related to latency issues on account of servers being located in the US. Interestingly JIBO Inc. also pointed to “rapidly changing consumer-privacy laws” in countries outside the US as complicating the delivery of personal data services from US-based servers. Since starting work on Jibo the EU has seen various shifts in its privacy legislation landscape, with Safe Harbor nixed and now replaced by the EU-US Privacy Shield; while the region has also updated its GDPR directive — due to come into force in 2018 [1].
Despite all these contingencies, JIBO Inc. raised US$ 70.4M in 5 Rounds from 13 Investors and the most recent funding (13.1M Series B) has been achieved on November 10, 2016! Investors look very confident in the future of the company.
Now the features that were promoted in 2014 are for most of them outdated with respect to the competition – especially with big names like Amazon Echo and Google Home coming in.
If we look at the videos on Jibo Inc. web site, I got the feeling that team behind JIBO is focusing on one special topic in order to stay in the game: develop a robot which is able to develop its own personality and expresses a range of emotion based on information learned from family behaviors. The ultimate goal is that family members will feel like they talk to a mate
Jibo learns by listening and asking questions. Jibo uses machine learning, speech and facial recognition, and natural language processing to learn from its interactions with people.
Breazeal did say that her speech engineers are building natural-language models to allow Jibo to respond in an engaging manner. They’ve also given Jibo a unique voice, courtesy of a voice actor who recorded some 14,000 phrases. From those, a text-to-speech engine can generate millions of utterances. But she adds that the actual words the robot will say are only part of its response—it will also use body language as well as alter the tone of its voice to suggest happiness, sadness, and surprise [2].
Over time, Breazeal promises, Jibo will learn details about individual users, offering more personalized responses. And then, she adds, people will treat Jibo not just like another gadget but as “part of the family.” That’s her fondest dream, and it would be a lasting legacy to leave in robotics [2].
Jibo Inc. is clearly positioning itself in analyzing human behavior in order to provide cute and fancy social robot – and it is the nice part of the story - BUT on the other hand they will acquire a tremendous amount of (very sensitive) information regarding human behavior (what do they like, not like, habits, opinion about political/social issues…) in the family circle that can be used for other purposes (marketing in the best case) and many companies will pay a lot of money for it! JIBO Inc. is competing with big names such as Facebook, Amazon or Google.
The three other companies are not so ambitious – and actually they do not have the resources JIBO Inc. has either! They are committed to deliver a product with (not so well) defined functionalities to entertain their backers.
UBTECH an established company with the know-how
Let us have a look at UBTECH. Founded in 2012, UBTECH is one of the big players in the humanoid robot market from China. It launched a campaign on Indiegogo to promote its new humanoid robot ALPHA 2 which was amazingly successful because they raised US$ 1.4 Million on Nov 2015. But UBTECH is not a startup like Blue Frog Robotics or Ingen Dynamics Inc:
- They already developed humanoid robots as a reliable product (ALPHA 1), meaning that they have the know-how (R&D, mass-production, testing, support),
- They have huge resources in terms of staff and the company is already quite big (more than 200 employees),
- UBTECH received US$20 million series A+ round from Qiming Venture Partners in 2015, according to China Money Network, followed by an “undisclosed strategic investment” by Chinese voice recognition software maker Anhui USTC iFlytek Co. at the end of last year. In July 2016, the stakes went even higher on a second round of investments worth $100 million, led by CDH Investments, along with CITIC Securities and other unnamed investors, which values the start-up at $1 billion post-money [3].
Despite all these assets, UBTECH needed one year to finally create an improved version of their ALPHA1 humanoid robot and start shipping ALPHA 2 to his/her backers in December 2016.
InGen Dynamics and Blue Frog Robotics - new comers which are struggling to make their social robots happen
InGen Dynamics and Blue Frog Robotics are new comers and they want to create their own social robots. These companies are true startups with ideas, energy, dynamic people and some – but limited financial resources and "stars" from prestigious universities among their founders and/or shareholders. They did a successful crowdfunding campaign and now they have to deliver.
Blue Frog did its campaign in 2015, when the social robot topic was quite hot with few other projects to compete with.
InGen Dynamics is more recent (April 2016) and was innovative on the mobility issue by proposing a ballbot which is a dynamically-stable mobile robot designed to balance on a single spherical wheel (i.e., a ball). The first successful ballbot was developed in 2005 [4] by Prof. Ralph Hollis of the Robotics Institute at Carnegie Mellon University (CMU), Pittsburgh, USA and it was patented in 2010. It is quite challenging to commercialize a ballbot because the control of the balance is touchy but you can come up with a very graceful navigation (a good example is the video of Rezero [5] designed by Péter Fankhauser and his students at the Swiss federal institute of technology in Zurich in 2011).
Otherwise both AIDO and BUDDY are quite similar in “social” features they are offering to backers. And they are offering a lot of features which may not be fully implemented when they did the crowdfunding campaign.
Limit of crowdfunding platforms for new comers
Actually this section applies to the new startups without “stars” from prestigious universities among their founders and which are really startups without a real experience in mass-production.
If we look at comments from backers, it is obvious that these companies are facing difficulties to deliver. We can make reasonable guesses which are related to the way crowdfunding platforms are operating:
- Prototype: awesome videos show a prototype which is working for some few use-cases. However a prototype is not a product. A prototype demonstrates the main functionalities of the future product. Therefore bridging the gap between the prototype and the product requires a huge amount of work. Add more debugging and implementation of functionalities which have been announced/promoted in the campaign but not implemented in the prototype and you end up with an important delay in achieving a product fully functional. Social robots like AIDO and BUDDY are really promoting a lot of features and they have very limit;
- Hiring: if the campaign is successful you need to hire more people, not only to strengthen your R&D department BUT also new ones with really different profiles (sourcing, supply chain, testing, support, documentation…). You need to re-organize your company, develop and establish processes so that the different departments/units/teams talk to each other efficiently. It takes time and money which brings us the crucial point: MONEY;
- Money: Startups show very aggressive prices (with respect to the functionalities offered by the robot) in order to increase the number of bakers and reach the funding objective. At a first glance it looks good for the bakers because they have the feeling to buy a cute toy at a low price but the point is: this price cannot cover on one hand the R&D effort to go from the prototype to the product and on the other hand its mass-production. This is why, if the campaign is successful, the startup will have to achieve two competing goals in a very short period of time:
- First priority FIND MONEY by chasing venture capital funds – the more money the startup raised during the crowd funding campaign, the more the startup attract the attention of the Medias and thus the chance to involve venture capital firms;
- Make a product and Mass-product it. And here is the issue, mass-production means find the factory that will do it, negotiate the delay and contracts (these factories have other customers to serve), eventually (if the factory is not doing it) purchase components, molds (often the most expensive part) … And the number of units that will be produced – few hundreds to less than one thousand in most cases - is not big enough to negotiate a significant discount. As mentioned before, the price tag of the product is not high enough to cover the expenses. This is why rise money is crucial in order for the company to guarantee that the robot will be delivered to bakers.
- Legal issues: more and more countries are voting laws to protect the privacy of their citizens in the digital world and it may be conflicting with technologies used by social robots that need to send back videos and/or audios information to centralized servers in order to process the data (mainly related to speech to text, face/object recognition…) in order to compute a “reaction” which is sent back to the robot.
Crowdfunding platforms do not control the “maturity” of the prototype which is promoted by a given company. Most of the time, videos are made with a mockup that is able to achieve a limited set of functionalities, just the ones that will please the audience. In such condition, if the campaign is successful it is not a surprise to notice huge shifts in the planning that was announced during the campaign.
[1] https://techcrunch.com/2016/08/11/uh-oh-crowdfunded-social-robot-jibo-wont-now-ship-internationally/
[3] http://www.nanalyze.com/2016/12/humanoid-robots
[4] Tom Lauwers; George Kantor; Ralph Hollis (October 2005). "One is Enough!" (PDF). 12th International Symposium on Robotics Research.
[5] https://www.ted.com/talks/peter_fankhauser_meet_rezero_the_dancing_ballbot