iCub vs temi V3

1.05 m

1 m

33 kg

12 kg

2 cameras (Pointgrey Dragonfly 2, 640x480) in the eyes · 2 stereo omnidirectional microphones · inertial sensors (3-axis gyroscope, accelerometer and geomagnetic, BOSCH BNO055) · absolute magnetic and incremental joint encoders · 6 six-axis force/torque sensors plus 2 ankle sensors · more than 3000 capacitive tactile sensors (skin)

360 degree LIDAR · 2 depth cameras · RGB camera · IMU · 6 time-of-flight linear sensors · 4-mic omnidirectional array

2 x Pointgrey Dragonfly 2 (640x480) mounted in the eyes

13 MP autofocus RGB camera, 1080p at 30 FPS, 120 degree FOV · ToF depth camera, 30 FPS, up to 5 m, 90 degree FOV · 2 navigation depth cameras

Intel i7 7600 CPU, 4 GB RAM, 32GB SSD

Two ARM hexa-core computers (dual-core Cortex-A72 1.8GHz + quad-core Cortex-A53 1.4GHz each): one for human interaction, one for navigation

iCub software repository: kinematics and dynamics, force/impedance control, reaching and grasping behaviours, attention system, online object learning and recognition, speech interfaces, OpenCV, Point Cloud

ASR with far-field voice, NLP, speech-to-text and text-to-speech engines; Alexa and Hey temi assistants; self-learning navigation AI

Ubuntu LTS 64 bits

temi OS on Android 11 (interaction computer); Linux (navigation computer)

Yes

Yes

research

enterprise