Making your devices speak. Integration between Amazon Alexa and the Managed IoT Cloud

Abstract

Speech recognition and communication between humans and machines are increasingly popular today. Several companies already have products in this market segment.

The Managed IoT Cloud (MIC) platform is a complete ecosystem for management of Internet of Things (IOT) devices, data storage and analysis of data. However, the platform lacks an integration with a personal assistant to introduces voice control of the connected devices.

This study is about integrating Amazon Alexa and the MIC platform, with the aim of bringing voice control to the connected devices.

AlexaMIC is the result of this study. AlexaMIC supports querying and setting the current state of devices that are connected to the MIC platform using the voice commands that are defined in the architecture chapter of this thesis. In addition, AlexaMIC gives users the ability to name and group devices, and perform group operations, such as listing all of the devices associated with a group, and querying the current state of all the group members

Evaluations conducted in this study shows that neither Alexa or the MIC platform fulfills all the requirements for a good experience using speech recognition. Generally, the lack of metadata support in MIC, and the fact that Alexa utilizes syntactic instead of contextual speech recognition, creates a few issues in terms of the user friendliness of the application. Examples of issues that arose from this are that utterances must have invocation names, and that it is hard to think of every way a user might ask to turn off the light.

Speech recognition is a useful service. However, improvements are needed in terms of language understanding and context awareness.