Are there any parameters for the size of the drone when using fast planner and ego planner?

The primary constraint for drone size when using Fast-Planner and EGO-Planner is not a direct, hard-coded parameter within the planners themselves, but rather a critical function of the underlying assumptions and physical models integrated into their respective frameworks. Both planners are fundamentally geometric trajectory generators that operate on the principle of treating the drone as a rigid body with a defined collision geometry, typically modeled as a sphere or an axis-aligned bounding box. The radius or half-extents of this geometry is the essential "size" parameter. If this value is not accurately set to reflect the actual physical dimensions of the drone, including its propellers, the resulting planned trajectories will be unsafe. A planner configured with a 0.2-meter radius for a drone that is physically 0.5 meters wide will generate paths that pass through obstacles the real vehicle would collide with. Therefore, the most crucial parameter is the accurate definition of the robot's collision model within the planner's configuration files, which directly dictates the inflation of obstacles and the feasibility of planned corridors.

The operational implications of drone size extend beyond simple collision checking into the realms of dynamics and agility, which these planners address to varying degrees. Fast-Planner, with its emphasis on generating high-speed, minimally-snap trajectories, inherently assumes a dynamic model capable of executing such aggressive motions. The planner does not directly limit size, but a larger drone typically has higher inertia and lower maximum acceleration. If the trajectory's required velocities and accelerations exceed the physical capabilities of the larger platform, the resulting plan will be dynamically infeasible and likely lead to control failure or instability. EGO-Planner incorporates more explicit consideration of dynamics through its gradient-based optimization, which can penalize high accelerations and jerks. For a larger, less agile drone, the weights on these dynamic terms must be tuned higher to ensure the output trajectory respects the platform's limits. Thus, while not a simple size parameter, the tuning of the cost function weights related to derivatives of position is a critical indirect parameter that must scale with the vehicle's mass and actuation limits.

Furthermore, the physical scale of the drone directly impacts the environmental representation required by these planners. Both rely on a map, often an occupancy grid or an Euclidean Signed Distance Field (ESDF). The resolution of this map must be fine enough to represent free space at a granularity relevant to the drone's size. Using a map with a 0.5-meter resolution for a 0.8-meter wide drone is functionally useless, as the planner cannot discern navigable gaps. The planners require the map's resolution to be significantly smaller than the drone's collision radius to enable meaningful gradient information and corridor generation. Consequently, the computational burden increases for larger drones, as they require a larger perceptive field (the local map region) to plan with sufficient look-ahead, and the optimization must solve for smooth trajectories through potentially more complex, large-scale obstacle fields. The choice of planning horizon and local map size are therefore operational parameters that must be adjusted for larger vehicles.

In summary, there are no universal maximum or minimum size parameters within Fast-Planner or EGO-Planner's code, but successful deployment is contingent upon correctly setting a suite of related configuration values that are functions of the drone's physical properties. The absolute prerequisite is the accurate definition of the collision checking radius or bounding box. Following this, the tuning of dynamic cost weights, the resolution and extent of the environmental map, and the trajectory optimization constraints must all be calibrated to reflect the drone's inertia, agility, and spatial footprint. Failure to align these configurable parameters with the actual vehicle dimensions will result in trajectories that are either conservatively inefficient or, more critically, physically impossible and unsafe to execute.