Autodesk: AVA Video chat-bot

AVA (Autodesk Virtual Assistant) is the video chat-bot from Autodesk with a realistic human appearance and character.

A system is based on machine learning and recognition of a natural language, the knowledge base of the chat-bot is filled with records of chats, posts at forums, a key word, syntax, idioms, etc. It helps it to understand an essence of requests better. According to developers, over time AVA studies and becomes quicker and more smart.

Prototype of the AVA virtual assistant, "digital person" version

2017: Announcement

The developer company of technologies for 3D - design of Autodesk announced on November 20, 2017 the next generation of the AVA virtual assistant which not only will get realistic appearance, but also will be able to read out nonverbal signals of the interlocutor and tone of its voice. It is planned that the upgraded AVA will work in technical support for users of Autodesk.

The company began to work on the virtual assistant more than a year ago. At first it was the simple text chat-bot responding to the standard requests of clients. AVA processed more than 35 thousand requests a month, quickly solved standard situations and if could not cope itself — collected necessary information and sent to the responsible.

The majority of bots with which we interact today will not be recognized by emotional signals which we, people, perceive during the conversation absolutely instinctively and naturally. They can interact with us intellectually, but not emotionally. Autodesk works with Soul Machines over creation of the next version of AVA which will have a human face, character, a voice, capability to read emotional signals and to react to them — told in Autodesk. — AVA will be able to recognize nonverbal signals of the person that it is unavailable to traditional text chat-bots and voice assistants. Thus, it will provide more natural interaction interface.

Current version of AVA (at the left) and prototype of the "digital person" version

According to the statement of Autodesk, the next generation of AVA will have:

• The realistic human face (video is lower) designed by the director of Soul Machines Mark Sagar;
• Connectivities of a voice and video by the user. It will provide the deep personalized impression that AVA will distinguish, interpret and react to the emotions distinguished on a look, tone of a voice, etc. It is supposed that AVA will also be able to react to specific scenarios and to transfer the same signal types (emotions and tone of a voice respectively), reducing risk of the wrong interpretation;
• Users will be able to select whether they want to communicate the text, audio and/or video — as with the normal staff of technical support.

About process of creation of AVA of the next generation, on November 14, 2017

Robotics

You watch also (messengers)