Qdrant Edge Memory Robot
A companion robot project for the DeepLearning.AI short course on building AI assistants with on-device memory, using a camera and microphone to recognize objects, learn their names by voice, and recall when and where it saw them, entirely on the device.
This repository is the physical continuation of the DeepLearning.AI short course "Building AI Assistants with On-Device Memory," built in partnership with Qdrant. The course notebooks build the memory layer; this project connects that layer to a camera, microphone, object detector, and browser interface running on a device such as an NVIDIA Jetson.
Holding a TEACH button and naming an object stores it along with the place and time; showing it again logs a new sighting, and holding ASK lets a user query when an object was last seen. Object detection and tracking (YOLOE) produce a CLIP image vector, speech recognition (Whisper) produces a name, and a Nomic text vector links the two in a local Qdrant Edge shard, with detection, speech, embedding, and search all running on the device with no cloud service or API key.
A computer with a webcam and microphone is enough to explore the code without building the physical enclosure; Python 3.12+, uv, and Windows, macOS, or Linux are required.