Hi all,
I’m working on SeekSense, a semantic search API that lets mobile robots find objects/locations by name in unfamiliar environments, without per-site training.
Rough idea:
• Robot streams RGB(-D) + pose
• SeekSense builds a language-conditioned map on the fly
• You call an API to get “next waypoint” suggestions and, when the target is visible, an approach pose
You keep your existing stack (ROS / ROS 2 / Nav2, etc.). We just handle the semantic side: “What should I look for, and where should I go next for that?”.
Examples of the kind of behaviour we care about:
• “Find the cleaning cart for ward C”
• “Locate pallet 18B near goods-out”
• “Go to the visitor kiosk in atrium B”
I’m looking for a small number of teams to join an early alpha:
• AMRs or mobile bases in warehouses, depots, hospitals, or labs
• Ideally, ROS / ROS 2 / Nav2 is already running
• Real “go find X” or recovery tasks you’d like to automate
What I’m offering:
• Free early access to the API
• Hands-on help wiring it into your stack
• Flexibility to adapt the API to what you actually need
If you’re interested, there’s a short overview and sign-up form here:
Happy to answer questions here or share more technical details if that’s useful.