Siri, Cortana, Google Home or Alexa – all household names in our Smart Homes. However, Felix Eichert, Product Owner for Connected Car projects, asked himself if these smart voice assistants also work in places where hands-free commucation makes particular sense: in our cars. In this interview he talks about the integration process and how it feels like to breathe life into the buzzword “innovation”.
You and your team have integrated Alexa into the car. Where do you start with something like that?
My first encounter with Alexa was in 2016 at the JavaOne in San Francisco, where it was introduced for the first time to the general public. I had the feeling that this topic would become interesting for us since it made voice connection – not an easy topic in the automotive industry – usable for consumer and elevated it onto the next higher level. And since it really made quite an impression with me I used the breaks between speeches to familiarize myself with the technology. Back home I set up the first skill that allowed to output at least a simple text.
Alexa was not on the market yet back then. How did you get started?
At first I used a simulator, i.e. an application that recreates the functionalities. I was one of only a few privileged people who were permitted to attend the introduction of the closed Beta Box in Germany and then allowed to buy it afterwards. After a few weeks I had a relaxed sit-down with a couple of my colleagues at one evening to wax philosophical about possible applications. We had the idea to connect Alexa with our ACON box to allow the request for vehicle information (e.g. location or car lock status). After I had introduced the idea to our Managing Director, he spontaneously made his own Box-equipped car available and took the afternoon off to test the functionality together with us.
The Proof-of-Concept was a success obviously: What to do with it next?
After setting up additional skills, everything went really fast. Since we work in close partnership with our cleints, I was able to introduce the project there fast and with no red tape involved. Currently the project is with our client, is ready for use and is being planned for future car generations.
When an idea becomes a client project: How independent is implementation then ?
First and foremost: We support good ideas intensively. Of course, we had an interest in the project eventually breaking even and potentially making money, but we had not been given any targets to meet. Quite the contrary, I experienced a lot of support , also from high up. Additionally, I was given the latest Amazon Echo devices from the company so I could present the client with something they could touch. Not every company would do that. The biggest help, however, were my colleagues who always came and offered new ideas. And because our team pools a lot of know-how that can be used in very different customer projects, we were able to explore ideas in many directions. Colleagues were available fast and at short notice – that’s just the mentality here. Same is true for exchanging ideas and expertise.
What does that look like in everyday situations?
On the one hand, you constantly profit from the knowledge your colleagues possess, complementing yours. On the other hand, we have the “Rack your Brain “ format. That is a regular time slot of about 2-3 hours to talk about technical issues. I, for example, used it in a workshop to demonstrate to my colleagues what the Amazon Alexa is capable of and explain the tooling. A number of them added some of their own skills in the context of this workshop. And this is how it grows naturally.
In a nutshell: Could you summarize the set-up used?
The implemented technologies are the client’s central platforms: AWS and the AWS Lambda server landscape. The database was a mySQL-data base and for programming language I used Java 8 with the usual frameworks and also the Lambda server landscape, since it facilitates running skill logic. That makes setting-up your own server moot. You just input the code via Lambda and let it run. Plus, you don’t have to worry about on which and how many servers the code is run. Sounds more complicated than it is.