2025
Research foundations
Vision-language-action models, reinforcement learning, robot learning, navigation and agent design.
Progress
The roadmap records outcomes. It is not a weekly engineering log.
Playable Patricia Town, robot demos and partner-defined tasks.
Vision-language-action models, reinforcement learning, robot learning, navigation and agent design.
Conversation, memory, procedures, controlled actions and persistent activity state.
Patricia components running on a real mobile prototype.
Playable loop, scenario editor, end-to-end robot demo and phone/web continuity.
Integrate the next robot and test one bounded task with a care or domestic partner.
A repeatable task, clear support model and evidence of value.