What is Pepper

[Robot] Introduced Pepper It is a tablet-equipped robot released by Softbank.

I will summarize what you can do and what kind of utilization you can think of. How can I build a system unique to robots? We are always looking for ideas, so don't hesitate! (Lol)

What kind of utilization will create an unprecedented solution?

Since my work is medical, I am involved, so it will be better than that To be honest, how widespread Pepper will be depends on the medical and welfare field.

** ・ Utilize simple operability ** I think it is a great merit in the medical and welfare field that you can use various functions just by looking at your eyes and calling out.

** ・ Utilize the tablet integrated type ** By using a tablet, I think it can be used with a higher degree of control.

** ・ Utilize various sensors **

** ・ Take advantage of being mobile ** I don't know how far it is possible, but I think there is a new way to use it by going around the facility and acquiring information.

** ・ What kind of disadvantages can be considered ** I would like to verify the disadvantages while actually using it.

Choregraphe box function

Choregraphe is a tool for developing a system that runs on Pepper, and you can do various things without programming. The functions are organized as a box, and you can let Pepper do various things just by connecting the settings and flow of it. You can see what can be easily achieved from the functions of the box, so we will summarize them by function.

Microphone ・ Word recognition function Register the word you want to read and recognize it as a word if it exceeds the set threshold. → If the question has a fixed answer, various functions can be considered with this alone. ・ Know the position of the sound source

** Camera face recognition function ** ・ Understand the number of faces in the camera ・ Learn the face being photographed ・ Determine age ・ Determine gender ・ Judge smiles and emotions ・ Chasing the face ・ Track a specific object → Place a mark as a mark in the facility and move it I'm wondering if it's possible to put something like masking tape on the floor and move along that line. ・ Image recognition → Is it possible to recognize changes in the facility? ・ By learning the object you want to recognize the image (specify the target from Choregraphe), it becomes possible to recognize it.

touch ・ Detects the stimulus of the bumper part (it can be judged whether it is on the right side or the left side) ・ Detects the contact of the foot part (probably the front and back can be judged) ・ Detects contact with the head (front, middle, back) ・ Detects contact with the hand (one place) ・ Get tablet touch events

** Other actions ** ·Speak ・ Move (various animations are available) ・ Play audio ·take a picture ・ Infrared communication (Synchronous processing of multiple robots can be performed) ・ Obstacle detection with sensor (determination of right side or left side) ・ Detects that a person has entered or exited the defined area ・ LED control ・ Send and receive emails ・ Perform tablet-related operations

** Functions that can be realized (with programming) ** ・ Record voice ・ Recognize the desired action (blinking, snapping, etc.) ・ Link with apps created for tablets

Sensors

About the actual sensor from the material Pepper Product Specifications Head: Microphone x 4, RGB camera x 2, 3D sensor x 1, touch sensor x 3 Chest: Gyro sensor x 1 Hand: Touch sensor x 2 Legs: Sonar sensor x 2, Laser sensor x 6, Bumper sensor x 3, Gyro sensor x 1, Infrared sensor x 2 3D camera that grasps the surrounding environment Ultrasonic system that can avoid obstacles and move safely

Touch sensor

3 heads and 2 backs

From the official instruction manual image Looking at the hardware specifications, it seems that the strength of the touch cannot be determined.

Switches are simple sensors that return a 2-state float value 0.0 (unpressed) or 1.0 (pressed). Beware that the value is read only every 10ms, and very fast change may be unseen.

Visual / objective sensor

Head camera, 3D sensor and leg bumper, sonar, infrared, laser

From the official instruction manual image

There are two cameras, one above the eyes and one above and below the mouth.

Hearing / voice sensor

4 microphones on the head I concentrate on the top of my head.

Output (communication) method

speaker The speaker is like the ear, not the mouth. It's the best way to communicate with Pepper, isn't it?

gesture

** LED lamp **

From the official instruction manual image and Aldebaran documentation

There are 8 LED points on the eyes and 10 LED points on the ears, and it seems that complicated color expression is possible. Since there is no point notation on the shoulder part, it is probably an expression of only one color.

NAOqi sensor related API

Aldebaran NAOqi Sensors & LEDs Aldebaran Pepper - Actuator & Sensor list

Other references

[Aldebaran documentationo Pepper TechnicalOverView] (https://cloud.aldebaran-robotics.com/naoqidoc/2-3/family/pepper_technical/index_pep.html)

Other notes

In the sense that it is a camera that can shoot at all times, isn't it the turn of AR?

[Pepper] How to utilize it?