02 - Open source - active prototype
RoutePilot
Autonomous urban mission planning under real-world constraints.
An autonomous operations agent that turns natural-language errand goals into verified, time-and-budget-aware urban itineraries grounded in live mapping data.
Vision
The premise behind RoutePilot.
Navigation should understand the mission, not only the destination. RoutePilot coordinates opening hours, budgets, dwell times, priorities and disruptions into one executable plan, then adapts the remaining route as conditions change.
- Decomposes ambiguous goals into explicit stops and constraints
- Verifies places, opening hours, distances and travel times
- Sequences multi-stop missions around urgency and time windows
- Replans remaining legs when a stop closes or a delay occurs
Constraint-aware planning
RoutePilot treats a day of errands as an operational problem rather than a list of directions. Its agent loop perceives the goal, decomposes it, calls grounded tools, optimises the stop order and formulates a mission.
- Time-window and budget constraints
- Priority-aware stop sequencing
- Return-to-origin planning
Grounded and adaptive
Place and route data are checked against Google Maps services before a plan is presented. Mission telemetry supports continuous replanning without forcing the user to begin again.
- Place and geocoding verification
- Traffic-aware route calculations
- Mid-mission disruption handling