Bot Libre AI Applications

The Bot Libre platform can be implemented in many different artificial intelligence applications. In particular, Bot Libre has been used to create mobile and web applications by using their bots and deep learning analytics. The latest Bot Libre application being developed for mobile devices demonstrates how AI can be used.

In this application, users can explore a variety of different locations to chat with Bot Libre’s chatbots. One feature of this application is to be able to walk around the different scenes with the chatbot’s avatar following you. To do this the avatar must avoid the different obstacles in the scenes. One way to accomplish this is to use the built-in AI in the application to calculate the shortest path to the user. The shortest path is calculated by using a pathfinding algorithm and it only takes into account the walkable areas in the scenes.

VIDEO1

In addition, another feature of this application is for the chatbot’s avatar to wander randomly around the scenes. Similarly to the following feature, the avatar must avoid obstacles when wandering around. Again, this is accomplished using pathfinding AI.

VIDEO2

The one downside of the algorithms for the previous two features is that they need to have knowledge of the scenes to accomplish the pathfinding. Another approach to have the AI avatar move around the scene is to use Bot Libre’s deep learning analytics. Images can be passed to the analytics and the responses can be used to specify where the avatar should move. This approach is used in the application as an alternative way for the avatar to wander around the scene. An analytic is passed an image from the avatar’s perspective every second to determine if the avatar should continue walking forward or stop and change direction.

VIDEO3

Bot Libre’s NLP analytics are also used in this application to process the users’ chat messages. This allows the AI avatar to understand when it should do some sort of action. For example, if the user asked the bot to come to the bar in the bar scene, the bot could understand this and subsequently move to the bar. To determine if it has reached the bar or not, the application uses another image analytic to classify locations. Once the analytic is confident enough that the avatar has reached the bar, it will stop moving.

VIDEO4