1 of 1

Hot Pot Bot

Team E5: Isabel Basow, Christina Chi, Shane Oh

Product Pitch

Hot Pot Bot is a system that allows food to be cooked automatically while still maintaining the spirit of traditional hot pot, where friends and family sit around a table and drop raw ingredients including vegetables, tofu, and thinly sliced meat into a pot of boiling broth at the center, cooking and eating throughout the meal. With all the socialization and revelry going on, it’s easy to overcook and lose ingredients that are floating around in the broth.

We required a contactless way to distinguish ingredients and a reliable way to cook them evenly. To achieve this, Hot Pot Bot controls three motorized strainers that move ingredients into and out of the broth, and it uses computer vision to identify and track ingredients as they are dropped into the strainers to determine the proper cook time. Our system classifies the contents of a strainer with 100% accuracy and cook time is consistent within 2 seconds.

System Architecture

System Description

System Evaluation

Confusion Matrix over a test set of 30 ingredients

Overview Block Diagram

CV Frame with identified beef, bok choy, and mushroom

Overall System

Control

Tower

Strainers

Control Tower

Strainer Mechanism

Motors

Push Button

LED stick

Strainer Movement Tradeoff

Test Metrics and Results

Jetson Nano

Arduino Unos

Camera

Power Supply

Hot Pot

Buck Converter