Beyond flight capability
Stable take-off, cruise and return solve only part of an unmanned-aircraft mission. Inspection, firefighting, emergency response and low-altitude services also require the system to understand the task area, priority, restricted boundaries, terrain cost, wind and remaining range, then convert those conditions into an executable route.
The Autavia Type 7-3 planning study uses forest-fire inspection as its operating context. A mission departs from a base, covers a forest area within range and time limits, gives priority to observable positions in higher-risk areas, avoids restricted or high-cost regions, and preserves a return margin. The aircraft configuration provides vertical take-off and cruise capability; the mission layer determines where to fly, in what order and when to revise the plan.
Encoding the mission environment
The planning system separates the environment into computable layers. Fire-risk priority represents inspection value. Terrain slope and altitude contribute flight cost. Wind changes energy demand by direction. Restricted and obstacle regions create hard constraints. Range, endurance, speed and safe altitude describe the platform boundary. These layers form a unified planning input.
Coverage differs from point-to-point navigation. A shortest path does not guarantee adequate observation of high-value areas, nor does it balance coverage against the return reserve. The process must determine waypoint generation, visit order, obstacle avoidance and route smoothing.
Layered planning workflow
Risk-weighted waypoint generation allocates more observation points to higher-priority areas. A genetic algorithm with 2-opt refines the visiting order. An improved A* search resolves local obstacle regions. B-spline smoothing reduces abrupt changes in the final route. Each layer can be inspected independently and then combined into a complete plan.
This structure supports engineering diagnosis. Coverage issues lead back to waypoint generation. Excessive route length can be addressed through sequencing and constraints. Routes close to obstacles point to the cost map and search parameters. Sharp turns can be treated during smoothing. The mission system remains a set of interpretable modules.
From interface to mission loop
The Autavia Mission Agent prototype brings mission inputs, planning state and results into one interface. An operator can define the task area, set constraints, initiate planning, and inspect the route, risk distribution and status log. The current prototype demonstrates workflow and interaction logic. Future versions can use real terrain, weather, vehicle state and mission telemetry.
A complete loop also requires execution monitoring and dynamic replanning. Wind changes, degraded communications, temporary restrictions and energy use may require route updates. The plan must remain consistent with flight control, the ground station and mission payloads so that it becomes a live operational plan.
Direction of autonomous mission systems
Next work will focus on real terrain data, dynamic constraints, field missions and system interfaces. Mission intelligence gains value through a dependable engineering loop: environmental information enters the model, the model produces an executable plan, the platform reports state, and the system adapts the mission.
Moving from autonomous flight to autonomous missions requires the air vehicle, perception, planning, control and ground system to share one architecture. Autavia is developing along this path.