Automated voice assistants for interconnected devices

Published by Telefónica IoT Team IoT Connectivity Hub

In a growing, interconnected world, the way we interact with devices is evolving rapidly. Given this need, the interface has to adapt to the most natural human communication possible. As a result, voice assistants are in the spotlight of large projects and huge companies. In the future of the IoT, personal assistants will be a prime player. But first, many barriers have to be overcome to reach the perfect experience.

Automatic, smart assistants

Automated personal assistants, as defined by some experts, are devices that are able to respond to and meet the needs of users smartly. To do this, they need to be connected and able to handle information according to the specific time or situation. For example, they can organize tasks, arrange appointments, buy tickets or book a table in a restaurant, among many other things. They are also able to provide the user with accurate information. Today there are hundreds of automated, smart personal assistants able to solve our daily tasks. But with the development of devices and the emergence of new use cases, the interfaces need to be adapted to new situations.

Thus, automated personal assistants are rapidly evolving into voice assistants. This extends their utility and creates a much more natural and comfortable H2M interaction. Software like Siri, Google Assistant, Cortana and Alexa are the most widespread smart assistants, but far from the only ones. One of the most important innovations, which was added to all these types of assistants long ago, is the ability to learn from the interaction. This allows them to anticipate the user’s needs: when they will need a restaurant, if there are tickets available from their favourite theatre company, or when their plane will depart. Automatic personal voice assistants allow you to focus on other activities so you can use them while driving, running or doing other tasks.

This has also made it possible to implement these assistants natively, or by using the mobile phone, in other devices. For example, smart cars have their own automated assistant that allows them to interact with the vehicle by voice. Another key role of theses assistants will be in Smart Cities, where a lightweight, natural interface saves costs and increases the efficiency of any action. For this to be accomplished, it is essential to be connected to the network. Automated voice assistants cannot function properly if they are not connected to the Internet or interconnected with the other devices around them.

Behind the software

Automated assistants have been in full development for years. As technological techniques and potentialities advance, the possibilities they offer increase, too. Today, there are three main challenges facing these assistants. The first is related to the interaction between the user and the device: speech recognition. It is a purely technical matter but can lead to misinterpretations and misunderstandings of specific requests by the assistant, which makes it difficult to use. To solve this problem, Artificial Intelligence is using advanced machine learning techniques that allow the assistant to learn with each new interaction.

The second challenge, misinterpretation, is almost exclusively related to current AI techniques. There are several studies that show that the answers provided by automated assistants are unsatisfactory in certain cases. At times, communication with them is based on certain commands and overly specific orders, which makes the relationship with the device clumsy and inefficient. Again, the solution to this problem lies in the application of machine learning techniques that improve the assistants’ reaction so they can learn from the user’s previous behaviour.

Finally, the third big problem is the inevitable need for connectivity. Assistants need to access databases immediately. They also need to have an instantaneous connection to fulfil a task or perform a service. In an interconnected world, assistants can make the most of the information they work with, but without this availability, automated assistants lose virtually all their functions. Fortunately, the solution to this issue is the emergence of new IoT technologies, including LPWA, improvements in satellite connections and cellular network extensions.

In addition to the technical challenges, as automated assistants become entrenched in society, new issues associated with their use appear. Questions like the type of voice that it should have (and its social implications), what their purpose is, and which device they are associated with (a connected car, a smart refrigerator, a wearable) are defining the evolution of these assistants. For now, not only are there a series of technical challenges with solutions just around the corner, there are also usability issues that will need to be addressed. In any case, automated assistants are extremely promising. Thanks to this interface, the way we interact with devices associated with the IoT will be much richer and more natural, making these assistants an invaluable tool in an interconnected world.

POST EN ESPAÑOL

Telefónica IoT Team