Adaptive navigation
Adaptive navigation connects what a system perceives with how it moves. Detection, scene understanding, and agentic reasoning feed system commands, while feedback informs the next decision. Target following, investigation waypoints, and obstacle avoidance make the route responsive to observations gathered during the mission.
How it connects
- Perceive
- Understand the scene
- Decide
- Send commands
- Adapt from feedback
Follow a target with defined separation
Predictive tracking connects an object’s estimated movement to the system’s next motion. A configurable standoff distance defines how far the system should remain from its target, while reacquisition handles a target that drops out of the current track. The navigation profiles described in the presentation cover people, vehicles, and assets.
Investigate and return to the route
When perception detects an anomaly during a mission, navigation can insert investigation waypoints into the active route and return to the original route afterward. This behavior sits within a repeating sequence of perception, scene understanding, reasoning, commands, and feedback. New observations therefore inform movement while the mission is underway, rather than only during initial planning.
Adjust clearance to the moving system
Depth estimation and sensor fusion combine observations into a three-dimensional occupancy grid: a spatial representation of where obstacles occupy the environment. Avoidance uses safety margins that scale with velocity, linking clearance to how the system is moving. The resulting response feeds the same perception-to-command loop used for target following and route adaptation.