Furhat

A Furhat robot head sits on a desk. It has a white base and a back projected face.
Ready for face to face conversations. Photo: Furhat Robotics

Furhat is a conversational robot head with a back-projected face designed to interact with humans in social settings. With highly expressive facial animations and a vast library of voices, it is used in academic research, education, and commercial innovation around the world.

Creator

Furhat Robotics

Year
2018
Country
Sweden 🇾đŸ‡Ș
Categories
Features

Rate this Robot

Overall Rating

Would you want this robot?

Appearance

Neutral

Did you know?

The name Furhat comes from a fur hat that was put on the first robot prototype, which had wires sticking out of the back of its head and the team needed to hide them during an event.

A man sits programming at a computer sitting next to a tabletop Furhat robotic head displaying a male face.
Projection technology and a Unity animation engine allow for customizable faces. Photo: Furhat Robotics
A smiling boy lies on the floor interacting with a Furhat social robot, which has a glowing backlit projected face.
You light up my life. Photo: Furhat Robotics

History

Furhat Robotics began as a research project at Sweden's Royal Institute of Technology (KTH), where a team was exploring ways for machines to hold more natural, face-to-face conversations. In 2012, an early prototype—a robotic head with a projected, animated face—demonstrated how combining speech synthesis, computer vision, and facial animation could make interactions feel less mechanical. The work attracted support from KTH Innovation, and by 2014, researchers Samer Al Moubayed, Jonas Beskow, Gabriel Skantze, and Preben Wik had spun it out into a company headquartered in Stockholm.

From its early years, Furhat Robotics positioned itself as a platform company, providing hardware, software, and tools for others to develop their own social robotics applications. In 2018, the company introduced a new version of the Furhat robot, a stationary head with a back-projected face that can switch between different characters, genders, and expressions. Motorized head movements, eye-tracking, and conversational AI make it capable of nuanced, adaptable interactions. Developers access the system through an SDK and simulation tools, integrating it with AI services and external applications.

In 2022, Furhat expanded its hardware expertise by acquiring U.S.-based Misty Robotics, known for its small mobile social robot, Misty II. The move broadened Furhat's portfolio and aimed to accelerate development of next-generation social robots for use in education, healthcare, public services, and human-robot interaction research.

A white base holds a projected robot head that wears a fur hat.
The first generation Furhat could often be found wearing a fur hat. Photo: Furhat Robotics
Three tabletop Furhat robots sit in a row. Each consists of a black base with a unique, back projected face.
These Furhats have no hats, but many faces, and can speak in 120 languages and dialects. Photo: Furhat Robotics

Previous Versions

The original Furhat robot head is seen from front and side view. It looks like a white box with a head shape held above it with some exposed electronics.
2018

Furhat

The first generation of social robot Furhat was released in 2018.

Specs

Overview

Voices can be customized to create robot characters with any ethnicity, age, gender, and personality. Face-tracking, multi-party interactions, external API integrations, data streaming, and logging tools.

Status

Ongoing

Year

2018

Website
Width
27 cm
Height
41 cm
Length
24 cm
Weight
3.5 kg
Sensors

Head with 1400:1 contrast 1280x720 pixel resolution, and 165 lumen projector. Main body with a fixed RGB 3.4 MP 120° diagonal FOV camera and 2 x 100 Hz-10 kHz digital, PDM stereoscopic digital MEMS omnidirectional microphones, set 180 mm apart on the robot's shoulders. Bottom RGB LED Ring for silver lining effect.

Actuators

3 x high speed servos, active feedback, 0.088° resolution, and 25 kg·cm stall torque.

Degrees of Freedom (DoF)
3 DoF (Neck: 3 DoF)
Materials

Injection molded plastic, polymer, steel rods, aluminum shaft, 3D-printed structural parts.

Compute

Intel Iris Plus 640 GPU and Intel Core i5 CPU with up to 3.40GHz, 8GB RAM and a 120GB SSD mass storage.

Software

FaceCore, AudioCore, CamCore, OpenVino CV, Unity, Ubuntu Linux OS, Kotlin for native programming, Java, Web-based user interface, Websocket API, Remote API for Python, Javascript, C++, Blockly, Google Cloud ASR, Microsoft Azure ASR and TTS, Microsoft Azure OpenAI, Amazon Polly TTS, Acapela TTS, ElevenLabs TTS.

Power

19V, 90W power input

Cost
US $28,000 – $34,000