The participant faces the camera; system calls out expressions one at a time (e.g. "Happy", "Surprised", "Angry"). Correctly matching the expression moves the bot one step forward on a straight track. Fastest bot to the finish line advances. Time frame: 2–3 min per participant.

• One facial expression prompt is shown at a time. Prompt order is randomized per participant.
• Each prompt allows exactly one attempt — no retries or re-dos.
• A fixed number of prompts per participant; each correct, clearly-detected match earns one point.
• Time limit per prompt: 5 seconds.
• Top-scoring participants (or those above the qualifying accuracy threshold) advance to Round 2.

Sudden-death, single expression call, fastest correct match wins.

Participant's bot starts at a fixed point on a maze grid displayed on screen. Each detected expression triggers the corresponding move. Participant must chain expressions in the right sequence to navigate walls and reach the finish tile fastest. If no one finishes within the time cap, furthest distance from start wins/first bot to reach wins.

• Each facial expression maps to one fixed directional action for the bot.
• A move is only registered when the expression is held clearly and detected with confidence; brief or accidental expressions do not trigger movement.
• Moving into a wall results in no movement; that turn is forfeited.
• First bot to reach the finish tile within the time cap wins.
• If no participant finishes, the bot with the greatest distance covered from the start tile wins.
• Backward movement is not permitted in this round.

Longest distance covered within the time cap; if still tied, sudden-death with first correct expression wins.