Inspiration What inspired us
Most AI agents are limited to browsers and software tasks. They can answer questions and call APIs, but they cannot move through a physical environment, change their point of view, or investigate something a camera cannot already see.
We wanted to bring an AI agent into real life by giving it mobility, vision, memory, and physical tools. This inspired myDEAR — My Digital Environment Assistant Robot, an autonomous assistant that helps people recover lost belongings and receive visual updates about their pets or babies.
What it does
When someone loses an item, they open the myDEAR web or mobile experience and describe what is missing. The agent asks useful follow-up questions about the item's color, shape, size, brand, identifying marks, and last-known location. It converts those answers into a structured visual description.
Before starting a physical search, the agent checks the lost-and-found database for reports and objects that the rover has already discovered. If a likely match exists, myDEAR can immediately present that record to the user.
If no match exists, the agent deploys the rover. The rover starts from its assigned position and autonomously searches the indoor environment. It moves, turns, changes camera angles, captures observations, and evaluates what it sees. The agent uses each observation to decide where to search next.
When the rover finds the requested object, it records the image and relevant details, updates the administrator dashboard, and notifies the user. If it encounters other potentially lost items during the mission, it records those as well. These discoveries make future searches faster because the agent checks the accumulated database before deploying the rover again.
myDEAR also supports pet and baby check-ins. A user can request an update through the web or mobile experience. The rover observes the requested area and sends the user a situation update with available visual evidence. These features assist caregivers but do not replace adult supervision, medical advice, or certified monitoring equipment.
We faced a lot of challenges fixing the hardware integration with the Raspberry PI to connect with the agents Ochestration.
Built With
- agentic
- gemini
- jac
- javascript
- llm
- python
- raspberry-pi
Log in or sign up for Devpost to join the conversation.