Pepper Robot Companion — Desktop Interface

A Django-based desktop interface for controlling and conversing with a Pepper humanoid robot over MQTT. Supports text chat, voice input, and direct movement/expression commands — all from a single browser-based UI.

---

Project Structure

---

Backend/
│
├── api/
│   ├── __pycache__/
│   ├── migrations/
│   ├── __init__.py
│   ├── admin.py
│   ├── apps.py
│   ├── models.py
│   ├── tests.py
│   ├── urls.py
│   └── views.py
│
├── app/
│   ├── __pycache__/
│   ├── migrations/
│   ├── static/
│   │   └── app/
│   │       ├── css/
│   │       │   └── style.css
│   │       ├── js/
│   │       │   └── script.js
│   │       └── favicon.ico
│   │
│   ├── templates/
│   │   └── app/
│   │       └── index.html
│   │
│   ├── __init__.py
│   ├── admin.py
│   ├── apps.py
│   ├── models.py
│   ├── tests.py
│   ├── urls.py
│   └── views.py
│
├── project/
│   ├── __pycache__/
│   ├── __init__.py
│   ├── asgi.py
│   ├── settings.py
│   ├── urls.py
│   └── wsgi.py
│
├── db.sqlite3
└── manage.py

---

How It Works

The browser loads a single HTML page served by the app app. All robot interaction is handled by the api app via fetch calls from script.js. The Django backend communicates with the physical Pepper robot through an MQTT broker (default: localhost:1883) on the topic Pepper/control.

Browser  ──fetch──▶  Django API  ──MQTT──▶  Pepper Robot
                         │
                    NLP / STT / TTS

---

API Endpoints

All endpoints live under /api/. Every POST request requires the Django CSRF token passed as the X-CSRFToken header (automatically handled by script.js).

---

GET /api/status/

Fetches the current robot state. Called automatically on page load and by the Refresh / Check MQTT buttons.

Response

{
  "ip": "192.168.1.10",
  "face": "idle",
  "command": "stand",
  "mqtt_connected": true
}

---

POST /api/quick/

Sends a direct movement + face command to the robot. Triggered by sidebar action buttons and command pills.

Request body

{
  "movement": "wave",
  "face": "happy"
}

Response

{ "ok": true }

The view publishes "wave,happy" to the Pepper8697803647/control8697803647 MQTT topic.

---

Available Movement Commands

stand, rest, forward, backward, left, right,
wave, dance, swim, point, pushup, bow, cute,
freaky, worm, shake, shrug, dead, crab, fight,
punch, kick, dizzy, fall, glitch

Available Face Expressions

idle, idle_blink, walk, rest, dance, wave,
happy, talk_happy, sad, talk_sad, angry, talk_angry,
surprised, talk_surprised, sleepy, talk_sleepy,
love, talk_love, excited, talk_excited,
confused, talk_confused, thinking, talk_thinking,
dead, point, shrug

---

Setup

Requirements

Django
paho-mqtt
SpeechRecognition (or equivalent STT library)
pyttsx3

Run

python manage.py runserver

Make sure the MQTT broker is running on localhost:1883 before starting the server. The robot status panel will show offline if the broker is unreachable.

Django settings

INSTALLED_APPS = [
    ...
    'app',
    'api',
]
 
STATIC_URL = '/static/'

Ensure django.middleware.csrf.CsrfViewMiddleware is in MIDDLEWARE — it is required for all POST endpoints.

---