New chatbot created in WirelessCar tutored student project

To tutor projects with telematics students is truly a great opportunity. Together during an internship period we are able to dive into exciting new ways to develop and implement new solutions, something that is of great importance to continuously push the boundaries. One example is a chatbot which allows humans to communicate with cars using natural language – a student project that has already been presented to one of our customers.

Jens (Application Software Architect): Hello there Emma and Katri! What brings you to WirelessCar?
Emma & Katri: We are students from Gothenburg Technical College, which has one of its locations not far from your office. Each year we have an internship period and this year we have spent 8 weeks at WirelessCar. Both of us had requested to do our internships with a company in the telematics space, and fortunately our wish came true!

On a high level, can you explain your project for the internship period?
We have done a proof of concept for a chatbot, which allows humans to communicate with cars using natural language. The user can talk to the chatbot on the phone or send it text messages via Facebook Messenger, Slack or SMS. The chatbot knows the state of the vehicle it represents and replies to the user using the same communication channel. You can for example ask where the car is parked, how full the tank is or if it is locked.

What technologies and services have you used for the chatbot?
We have mainly used Amazon Web Services (AWS), as it is already very popular at WirelessCar. Primarily it was the world of the Lex and Lambda services we made a deep dive into, which for us was new and confusing as we had no previous experience with those services. In the past weeks, we have learned a lot about AWS in general and programming in particular. AWS Lex is used for the natural language interaction with the user. AWS Lambda is more traditional programming in node.js, which was good for us as we had some previous JavaScript experience. We also used a couple of other services from AWS, such as DynamoDB, API Gateway and CloudWatch.

 

What was the most difficult part of the project?
Our biggest problem was probably to get the chatbot to initiate contact with the user. For example, we wanted the car to tell its owner when it started to be low on fuel. Surprisingly, Lex did not have support for this and we had to find a guide to do it with Lambda instead. That felt a bit like a workaround but I guess the important part is that we got it working in the end.

What general advice would you give to other students — for example, next year’s internship students — even if they do a completely different project?
Our best advice would be to seek out more information for anything that blocks your progress. Also, don’t give up just because the answer is not of the first page of results from Google. Sometimes you have to puzzle together many different pieces of information to get something to work for a particular use case. So in short our advice is to not give up just because it seems complicated.

If you would give any suggestions to your teachers at Gothenburg Technical College, which would those be?
I think it is very good for students to work on a project which they themselves find interesting. It is a lot easier to work on something that you enjoy, and you learn a lot more than what is explicitly required.

Finally, what did you think about the internship period at WirelessCar? What were the best and worst parts?
The best part was the freedom, and how the project was interesting and relevant to us. The worst part was probably that the vanilla tea ran out sometimes — I didn’t really know how to handle that situation. You should be more attentive to the supply of vanilla tea.

We will certainly monitor the vanilla tea supply more carefully. We are very happy to have had you here. Thank you very much, and good luck with your studies!

If you are interested in having your internship period with WirelessCar, please talk to your teachers and ask them to get in touch with us. We are continually running proof-of-concept projects in many different areas. For example, artificial intelligence and machine learning are areas which have evolved significantly in recent years and are accessible on a completely new level. Also, while IoT is already around us, it is reaching new levels of maturity as the expectations on embedded devices are rising for each year.

We look forward to the next student project!

 

Leave a Comment

Your email address will not be published. Required fields are marked *