Iterations
Cardboard prototype
The first wheeled chassis built out of cardboard around a rudimentary sensor suite. Validates the platform mechanically before committing to a permanent frame.
Digital twin simulation
A continuously updated 3D representation of the robot's pose and immediate environment, reconstructed from its depth camera and IMU in simulation. This is the foundation the rest of the stack reasons on top of.
Twin + vision-language reasoning
Layering a vision-language model on top of the digital twin so the robot can describe what it sees and reason about its surroundings in natural language.
Chassis design & build
Replacing the cardboard prototype with a more user-friendly, approachable 3D printed chassis. Applies the chat pipeline from the cardboard version.
Live twin & navigation overlays
Live twin reflects the navigation state in real time. Applies the navigation stack from the digital twin simulation.
Navigation: failure
The early end-to-end navigation runs: from chatting to navigation. Mostly useful for surfacing what was still broken. Wheels got stuck, camera connection unstable, slow model response times, etc.
Navigation: point-to-point
A clean point-to-point navigation indoors based on natural-language instruction. The robot reaches a goal pose using its own map and avoids obstacles along the way.
Navigation: multi-step tasks
Multi-step navigation tasks driven from natural-language instructions. The reasoning layer decomposes the request into navigation goals and dispatches them to the planner as a subagent.
Outdoor adventures, off the grid
The robot explores outdoors without access to internet, using only its onboard sensors and reasoning stack. The robot is able to reach a goal pose in an unstructured environment.
Same mind, new body
The robot moves about in its new body, on four stubby new legs. It was remodeled, 3D printed, and assembled to address durability concerns and enable a more modular configuration, supporting future upgrades and modifications.