An educational voice chatbot robot that answers programming and technical questions via Gemini-live-2.5-flash and a Node.js server, with a NodeMCU ESP8266 and MG995 servo moving its arm in sync.
A project to develop an educational voice chatbot robot that acts as a technical assistant for students in programming, computer engineering, and networking, carried out by a team of students with technical assistance from Techno Enjaz. The robot takes the user's spoken question through a web interface, converts it to text, and sends it to Google's Gemini-live-2.5-flash AI model to generate a contextual answer that is shown as text and spoken aloud, while a servo motor moves the robot's arm in sync with the reply.
The project brings three fields together in one system: generative AI and natural language processing (NLP), real-time web technologies through a Node.js server and the WebSocket protocol, and IoT and embedded systems through a NodeMCU ESP8266 board and an MG995 servo. Practical tests showed working voice conversations and arm motion synchronized with the reply, although the report gives no numerical measurements of answer accuracy or response time.
| Item | Details |
|---|---|
| Project type | Educational voice chatbot robot with physical interaction |
| Field | AI in education, natural language processing, Internet of Things |
| Project status | Working physical prototype tested in live voice conversations |
| Techno Enjaz role | Technical support and assistance to the students during project development |
| Core technologies | Gemini-live-2.5-flash (Gemini Live API), JavaScript, Node.js, WebSocket, NodeMCU ESP8266, MG995 servo |
| Outputs | Voice and text conversation, spoken replies (TTS), synchronized arm motion, web interface with a debug log |
Students in technical fields need quick answers to programming and technical questions outside lectures and office hours, while instructors spend time answering repeated, common questions. The project belongs to the field of AI in Education (AIEd) and aims to provide a digital assistant available around the clock that understands the context of a question, not just its keywords, and encourages self-learning in a setting where students can ask without embarrassment.
To make the interaction feel more real, the project went beyond a software chat app and gave the robot a physical body with a moving arm and a speaker.
The project was carried out academically by the students, with technical assistance and support from Techno Enjaz during development. It is presented among the office's work as an example of its support for projects that connect cloud language models with electronic hardware and the Internet of Things.
The system runs on two synchronized paths that meet at the central server:
| Block | Role |
|---|---|
| User interface (Web Client) | Captures speech and converts it to text; displays the reply and converts it to speech |
| Central server (Node.js Backend) | Manages requests, organizes data flow, and synchronizes the robot's state |
| AI service (Google Gemini API) | Processes text and generates answers with the language model |
| Electronics (ESP8266) | Controls the servo motor and receives on/off commands |
The interface uses WebSocket with the server for real-time data transfer, the server talks to Google's servers over HTTPS, and the controller connects to the server via HTTP polling. The Gemini API offers two modes: a REST API for standard JSON requests and a Live API based on WebSocket for low-latency two-way communication, with settings such as Temperature, Top-K, and Top-P to control the nature of the output.
| Component | Role and documented specifications |
|---|---|
| NodeMCU ESP8266 | Wi-Fi connectivity (802.11 b/g/n) and servo control; L106 32-bit processor at 80–160 MHz, 3.3V operation |
| TowerPro MG995 servo | Moves the robot's arm; digital servo with metal gears, 4.8–7.2V, 10–12 kg/cm torque, rotation up to 180 degrees |
| 18650 lithium-ion batteries | Power source; 3.6–3.7V nominal, 4.2V fully charged |
| LM2596 voltage regulator | Steps voltage down to a stable level for the circuits; 4.5–40V input, adjustable 1.23–37V output, up to 3A |
| Battery management system (BMS) | Protection against overcharge, deep discharge, overcurrent, and short circuits |
| AC-to-DC switching adapter | Converts AC mains to DC to power the circuits |
| Speaker | Outputs the spoken replies |
Power flows from the source to the BMS for protection, then to the LM2596 regulator, and the servo connects to the ESP8266 through power and signal lines with a common ground shared by all components. The outer body is made of lightweight parts that hide the electronics while keeping them easy to reach for maintenance.
A simple web interface was designed for voice conversations and real-time system monitoring. It includes:
Practical tests of the prototype showed:
These are functional results for a prototype. The report gives no quantitative figures for answer accuracy, response time, or the number of questions tested, so no unverified accuracy figures are attributed to the system.
According to the directions set out in the project:
If you are working on an AI-powered voice chatbot or intelligent assistant built on language models and need technical help developing the idea or building a prototype, you can contact the Techno Enjaz office to discuss your project's requirements and the appropriate scope of support.
An educational voice chatbot robot that acts as a technical assistant for students in programming and technical subjects, built by a team of students with technical assistance from Techno Enjaz. It takes spoken questions and replies in text and voice using the Gemini model, moving its arm with a servo while it answers.
The web interface converts the user's speech to text and sends it over WebSocket to a Node.js server, which passes it to the Gemini API over HTTPS. The reply streams back, is displayed as text and spoken via Text-to-Speech, while the ESP8266 board polls the server over HTTP to move the arm in sync.
Software: JavaScript, Node.js, WebSocket, and the Gemini-live-2.5-flash model with speech-to-text and text-to-speech. Hardware: a NodeMCU ESP8266 board, a TowerPro MG995 servo, 18650 lithium batteries with an LM2596 regulator and a BMS circuit, an AC-to-DC adapter, and a speaker.
Tests showed working voice conversations in the browser, stable reception of Gemini replies and their conversion to speech, replies synchronized with arm motion, and stable servo and power operation. These are functional prototype results; the report gives no numerical measurements of answer accuracy or response time.
The robot depends entirely on the internet and cloud processing and cannot work offline, and its motion is limited to one arm. It uses no camera or face recognition and gives no medical, legal, or psychological advice; proposed improvements include conversation memory, local models, and multi-dialect support.
Reach out to Techno Enjaz engineering team to discuss technical feasibility, prototype hardware, and development scope.
Discover more engineering prototypes and academic implementations
A children's learning system that uses a camera to recognize animals with YOLO11, colors with HSV, and numbers with OCR, then shows and speaks the result in Arabic.
A computer vision system that detects knives and pistols in real time from a laptop or phone camera using two trained YOLOv8 models, Python, and OpenCV.
An interactive learning system for children that recognizes objects via the camera with YOLOv8, speaks their names, detects colors with HSV, and labels uploaded images.
