Mobile guide robot that answers visitors' questions by voice

The problem

A new visitor arrives at the faculty without knowing where to go, so reception staff repeat the same answer dozens of times a day. More importantly, there's nobody to ask outside working hours.

How it works

1. ListeningThe microphone picks up the visitor's question, and the system converts it to text.
2. UnderstandingThe text is matched against the faculty's knowledge base to work out what the question means.
3. ReplyThe answer is spoken aloud and shown on the screen at the same time.
4. MovementLaser scanning maps the space, and the robot moves around while avoiding obstacles.

Project features

• The robot combines four ways to interact: voice, touch screen, face recognition, and movement.
• It avoids obstacles and people using laser scanning, not single-point distance sensors.
• It relies on a local knowledge base specific to the faculty, so answers are precise and specific rather than generic.
• It works outside office hours, continuously.

Expected challenges

These are the main obstacles to expect during the build, laid out clearly so you know them before you start.

• ⚠ The full Raspberry Pi board isn't available from us, and language processing and face recognition need a stronger processor than microcontrollers.
• Noise at the faculty entrance hurts speech recognition accuracy, so the microphone's type and position matter more than the model.
• Navigating among moving people is much harder than avoiding a fixed wall.
• Face recognition raises privacy considerations and requires consent and clear disclosure of what data is kept.

The consultation gives you a solution for each one.