Affective Tutoring Robot for Arithmetic Retention

ISEF Category: Robotics and Intelligent Machines

Subcategory: Cognitive Systems  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A robot face can do more than blink and smile. If it reacts to your voice, it may change how you remember what you just learned. That makes this project part robotics, part learning science, and part AI. You will test whether emotional feedback helps students hold onto arithmetic skills.

What Is It?

This project asks a simple question with a smart setup, does a robot tutor help you remember math better when its face reacts to your mood? Your robot can use speech recognition to turn spoken answers into text, then a small language model can label the speech as confident, confused, frustrated, or neutral. The robot then changes its eyebrows and eyes to match that label. Think of it like a teacher who nods when you are on track and softens its expression when you look stuck.

The key idea is affective expression. Affective means emotional. Instead of a flat robot face that always looks the same, your system changes its face in response to student speech. You then compare learning retention between the expressive version and a flat-affect baseline. Retention means how much the student still remembers later, not just right after the lesson.

Why This Is a Good Topic

This is a strong science fair topic because you can test a real human-robot interaction question with measurable outcomes. You can compare two conditions, expressive and flat, then measure learning with the same arithmetic task and the same retention test. The project connects to assistive tutoring, classroom engagement, and AI-driven interfaces. You can learn speech-to-text, sentiment classification, experimental design, and basic statistics in one project.

Research Questions

• How does an expressive robot face affect arithmetic retention compared with a flat-affect robot face??
• What is the effect of sentiment-matched eyebrow movement on immediate math accuracy??
• Does a robot that responds to confused speech improve delayed recall of arithmetic facts??
• To what extent do students report higher engagement when the robot mirrors their speech sentiment??
• Which facial signal, eyebrow angle or eye shape, best predicts student recall after tutoring??
• How does the accuracy of sentiment labeling from speech affect the learning outcome??

Basic Materials

• Servo motors for eyebrows.
• OLED display for eyes.
• Microcontroller such as Arduino or Raspberry Pi Pico.
• Laptop or desktop computer with microphone input.
• USB microphone or headset microphone.
• 3D-printed or cardboard robot face frame.
• Jumper wires and breadboard.
• Power supply matched to the servo setup.
• Arithmetic tutoring prompts and answer sheets.
• Consent forms and assent forms for human subjects.

Advanced Materials

• Robot head frame with mounting points for servos and display.
• High-torque micro servos or micro linear actuators for facial motion.
• OLED, LCD, or small display panel for eye animation.
• Microphone array or higher-quality USB microphone.
• Laptop or workstation for Whisper and language model inference.
• Local model hosting setup or API access approved by your school and review board.
• Video recording setup for coding facial expression timing.
• Data logging system for speech labels, robot expressions, and response times.
• Statistical analysis plan with repeated-measures tests.
• Human subjects approval documents and debrief materials.

Software & Tools

• Whisper: Transcribes student speech into text for sentiment analysis and response logging.
• Python: Runs the robot control script, data cleaning, and statistical analysis.
• ImageJ: Helps inspect frame-by-frame facial motion if you record the robot face on video.
• Google Sheets: Organizes trial results, scores, and condition order before deeper analysis.
• R: Supports repeated-measures tests and graphs for retention and engagement data.

Experiment Steps

1. Define the tutoring task, the age group, and the exact retention measure you will compare.
2. Choose one expressive face design and one flat baseline so you can isolate the effect of emotion.
3. Plan the speech pipeline from microphone input to transcription, sentiment label, and facial response.
4. Build a within-subject schedule that balances order effects, so each student experiences both conditions fairly.
5. Decide how you will score learning, engagement, and delayed recall with the same rubric each time.
6. Set up a clean analysis plan before collecting data, including how you will compare the two conditions.

Common Pitfalls

• Training the sentiment model on classroom speech, which can misread short math answers as emotion signals.
• Changing the robot’s expression too often, which distracts students from the arithmetic task.
• Using different problem sets across conditions, which confounds learning gains with task difficulty.
• Forgetting to counterbalance condition order, which makes practice effects look like robot effects.
• Measuring only immediate accuracy, which misses the real retention question after tutoring.

What Makes This Competitive

A competitive version of this project needs careful experimental control, not just a cool robot face. You should isolate expression timing, sentiment accuracy, and retention with a clean within-subject design. Strong projects also test whether certain facial cues work better than others, or whether the robot helps only specific students. A deeper analysis with effect sizes, confidence intervals, and order-effect checks can push the work well past a classroom demo.

Project Variations

• Test the same robot with fraction problems instead of arithmetic facts to see whether expressive feedback helps harder concepts more.
• Replace the small LLM sentiment labeler with a simpler keyword-based classifier and compare learning outcomes across the two approaches.
• Compare child speech, adult speech, and scripted text input to see whether the robot’s emotional response matters less when language is more formal.

Learn More

• NIH PubMed: Search for review articles on social robots, affective feedback, and learning retention in children.
• NASA Open Courseware: Look for free engineering design and human factors materials that help with system planning and testing.
• MIT OpenCourseWare: Search for introductory courses on artificial intelligence, speech processing, and experimental design.
• USGS Statistical Methods Resources: Find free guidance on hypothesis testing, repeated measures, and data interpretation.
• International Journal of Social Robotics: Search recent peer-reviewed studies on robot tutoring and emotional expression through your library or abstract pages.