Furhat™ is based on an award winning design that uses advanced facial animations to render a very high resolution and powerful facial dynamics on a mask that is shaped after the model projected on it. In this way, what makes Furhat’s design very simple, quiet, and low-weight, is also what makes it superior to anything that exists in the world today.
Furhat patent-pending design is also incredibly customizable. Furhat’s face is magnetically attached to the head, and can be manually detached and replaced with other models. By combining that, with a different projected facial animation, a different face, and a specific set of skills that are made for that mask, developers can build and customize their robots with almost no additional cost. This also opens up possibilities to easily and systematically study different facial geometries and designs for different applications, user groups, and cultures.
SIZE: 254 x 199 x 500 mm (10" x 7.8" x 19.7")
BOX: Powder coated aluminium
MASK: Magnetic clip-on, high-end polymer, cured for optimal back-projection and light diffusion
MOVEMENT: 36 DOF for the face (expandable), 2 DOF mechanical neck with high precision servos
COMPUTER: Intel NUC i5-4250U Haswell, 8GB RAM, 80GB SSD m.2, Windows 10
CONNECTIONS: 2 USB 3.0, ethernet port, mini-HDMI for external monitor
POWER: 12V DC (default) and battery (optional)
CONNECTIVITY: Wi-Fi and Gigabit Ethernet
PROJECTOR: LED, 100 Lumen
SOUND: 30W premium 8cm (3.3") speaker and amplifier
SOFTWARE: Web interface, gesture design tool, open source SDK
SENSORS: Built in support for Frauenhofer SHORE, Microsoft Kinect, Intel Realsense, SingStar Wireless Microphones, Cone Microphone, Nuance ASR, Microsoft ASR, Cereproc, Acapella, and Microsoft TTS.
Once connected to your network, Furhat can be accessed from a Web Browser Interface which gives you:
- Wizard-of-Oz interface
- Programming of “skills”
- Configuration of facial textures, gestures, etc.
- Configuration of peripherals and software components, such as microphones, speech recognition, speech synthesis, and both Microsoft Kinect and Intel Realsense cameras
- Logging of events and interactions
You can program Furhat in two different modes:
- Passive Mode: You control the speech, gestures and gaze of Furhat directly. This is suitable if you already have a dialogue system running on your machine and want to connect Furhat to use it as an output device.
- Live Mode: Furhat runs everything that is needed for interacting with him, including speech recognition, dialogue system, cameras, etc. You can upload “skills” (applications) to Furhat and run them directly.
Passive Mode: Direct control
If you want direct control over the basic functionality of Furhat (speech, gestures and gaze), you can easily do so from any programming language, by sending JSON messages over a TCP/IP socket connection.
Furhat also comes with a very simple Java API that does this for you:
Live Mode: Developing Skills
In Live Mode, you can develop “Skills” for Furhat, either in Eclipse on your computer (using the IrisTK framework) and upload them to Furhat, or directly in the Web Browser in an easy-to-use XML format:
In Live Mode, you can also run Furhat in a Wizard-of-Oz setup. Sitting at another location, you can see the world from Furhat’s perspective and control Furhat’s behaviour. You can also mix skills with Wizard control, to get a semi-autonomous behaviour. This is very useful when doing research or testing out new ideas.
You can customize textures and gestures directly in your Web Browser. You can also purchase new SAPI compliant speech synthesis voices and install them on your Furhat.