2025 — Foundation
Explore embodied intelligence
Research into vision–language–action models, reinforcement learning, robot learning, navigation and the relationship between language models and physical action.
Roadmap & demos
The roadmap shows meaningful stages, not a live engineering dashboard. It is updated when evidence changes the story.
Public demonstrations, Bumi integration and partner-led use-case selection.
2025 — Foundation
Research into vision–language–action models, reinforcement learning, robot learning, navigation and the relationship between language models and physical action.
2026 — Build
Agent architecture, procedures, persistent state and controlled actions moved from design into working code on XLeRobot 0.3.
Current focus
Prepare public demos, strengthen reliability, extend phone and web continuity, and integrate Patricia with the Noetix Bumi.
Next
Select bounded tasks, develop evaluation measures, test with intended settings and prepare a carefully supervised pilot.
Longer direction
Support additional embodiments and settings, build reusable procedure libraries and create a practical integration pathway for partners and developers.
This roadmap is directional rather than a promise of fixed dates. Updated August 2026; the next update will follow a meaningful demonstration or partnership milestone.
Demonstrations
The cards below reserve the locations for external video or project pages. They can be replaced without changing the site structure.
Patricia receives a request, navigates on XLeRobot, follows a bounded activity and reports the result.
Open video page Coming soonPatricia retains progress, follows a reviewable procedure and involves a person at a defined boundary.
Open video page Coming soonAn activity continues between robot, phone and web without losing its place or authority structure.
Open video page Coming soonProject pathways
Help move the roadmap