Having continuous conversations with Vector

Workaround one of the bothersome aspects of communicating with Vector

Wanted to pen a short post about one of the newest features of Wirepod which allows you to have continuous conversations with the Vector robot. One of the most bothersome aspects of Vector has been that you need to always wake it up (using Hey Vector or bressing the button on his head) to ask a question. This made having a continuous conversation with Vector a bit painful.

New Voice Request

Wirepod has a new feature to make this part a bit less painful. Using the concept of actions requested by Large Language Models (LLM) facilitating Vector’s Knowledge graph, Wirepod pulls off a new trick to enable you to have a continuous conversation with Vector. This feature allows an LLM to request for a new action of a new voice request at the end of answering a question. The LLM may or may not embed this request. But if it does, Wirepod intercepts the request and moves Vector to the ready state of listening to another question. This leaves the LLM to have the control of whether it wants to allow…