Attention: Here be dragons (unstable version)

This is the latest (unstable) version of this documentation, which may document features not available in or compatible with released stable versions of Redot.

Checking the stable version of the documentation...

NavigationAgent3D

Experimental: This class may be changed or removed in future versions.

Inherits: Node < Object

A 3D agent used to pathfind to a position while avoiding obstacles.

Description

A 3D agent used to pathfind to a position while avoiding static and dynamic obstacles. The calculation can be used by the parent node to dynamically move it along the path. Requires navigation data to work correctly.

Dynamic obstacles are avoided using RVO collision avoidance. Avoidance is computed before physics, so the pathfinding information can be used safely in the physics step.

Note: After setting the target_position property, the get_next_path_position method must be used once every physics frame to update the internal path logic of the navigation agent. The vector position it returns should be used as the next movement position for the agent's parent node.

Tutorials

Using NavigationAgents

Properties

bool	avoidance_enabled	`false`
int	avoidance_layers	`1`
int	avoidance_mask	`1`
float	avoidance_priority	`1.0`
bool	debug_enabled	`false`
Color	debug_path_custom_color	`Color(1, 1, 1, 1)`
float	debug_path_custom_point_size	`4.0`
bool	debug_use_custom	`false`
float	height	`1.0`
bool	keep_y_velocity	`true`
int	max_neighbors	`10`
float	max_speed	`10.0`
int	navigation_layers	`1`
float	neighbor_distance	`50.0`
float	path_desired_distance	`1.0`
float	path_height_offset	`0.0`
float	path_max_distance	`5.0`
BitField[PathMetadataFlags]	path_metadata_flags	`7`
PathPostProcessing	path_postprocessing	`0`
PathfindingAlgorithm	pathfinding_algorithm	`0`
float	radius	`0.5`
float	simplify_epsilon	`0.0`
bool	simplify_path	`false`
float	target_desired_distance	`1.0`
Vector3	target_position	`Vector3(0, 0, 0)`
float	time_horizon_agents	`1.0`
float	time_horizon_obstacles	`0.0`
bool	use_3d_avoidance	`false`
Vector3	velocity	`Vector3(0, 0, 0)`

Methods

float	distance_to_target() const
bool	get_avoidance_layer_value(layer_number: int) const
bool	get_avoidance_mask_value(mask_number: int) const
PackedVector3Array	get_current_navigation_path() const
int	get_current_navigation_path_index() const
NavigationPathQueryResult3D	get_current_navigation_result() const
Vector3	get_final_position()
bool	get_navigation_layer_value(layer_number: int) const
RID	get_navigation_map() const
Vector3	get_next_path_position()
RID	get_rid() const
bool	is_navigation_finished()
bool	is_target_reachable()
bool	is_target_reached() const
void	set_avoidance_layer_value(layer_number: int, value: bool)
void	set_avoidance_mask_value(mask_number: int, value: bool)
void	set_navigation_layer_value(layer_number: int, value: bool)
void	set_navigation_map(navigation_map: RID)
void	set_velocity_forced(velocity: Vector3)

Signals

link_reached(details: Dictionary) 🔗

Signals that the agent reached a navigation link. Emitted when the agent moves within path_desired_distance of the next position of the path when that position is a navigation link.

The details dictionary may contain the following keys depending on the value of path_metadata_flags:

position: The start position of the link that was reached.
type: Always NavigationPathQueryResult3D.PATH_SEGMENT_TYPE_LINK.
rid: The RID of the link.
owner: The object which manages the link (usually NavigationLink3D).
link_entry_position: If owner is available and the owner is a NavigationLink3D, it will contain the global position of the link's point the agent is entering.
link_exit_position: If owner is available and the owner is a NavigationLink3D, it will contain the global position of the link's point which the agent is exiting.

Signals that the agent's navigation has finished. If the target is reachable, navigation ends when the target is reached. If the target is unreachable, navigation ends when the last waypoint of the path is reached. This signal is emitted only once per loaded path.

This signal will be emitted just after target_reached when the target is reachable.

path_changed() 🔗

Emitted when the agent had to update the loaded path:

because path was previously empty.
because navigation map has changed.
because agent pushed further away from the current path segment than the path_max_distance.

target_reached() 🔗

Signals that the agent reached the target, i.e. the agent moved within target_desired_distance of the target_position. This signal is emitted only once per loaded path.

This signal will be emitted just before navigation_finished when the target is reachable.

It may not always be possible to reach the target but it should always be possible to reach the final position. See get_final_position.

velocity_computed(safe_velocity: Vector3) 🔗

Notifies when the collision avoidance velocity is calculated. Emitted every update as long as avoidance_enabled is true and the agent has a navigation map.

waypoint_reached(details: Dictionary) 🔗

Signals that the agent reached a waypoint. Emitted when the agent moves within path_desired_distance of the next position of the path.

The details dictionary may contain the following keys depending on the value of path_metadata_flags:

position: The position of the waypoint that was reached.
type: The type of navigation primitive (region or link) that contains this waypoint.
rid: The RID of the containing navigation primitive (region or link).
owner: The object which manages the containing navigation primitive (region or link).

Property Descriptions

bool avoidance_enabled = false 🔗

void set_avoidance_enabled(value: bool)
bool get_avoidance_enabled()

If true the agent is registered for an RVO avoidance callback on the NavigationServer3D. When velocity is set and the processing is completed a safe_velocity Vector3 is received with a signal connection to velocity_computed. Avoidance processing with many registered agents has a significant performance cost and should only be enabled on agents that currently require it.

int avoidance_layers = 1 🔗

void set_avoidance_layers(value: int)
int get_avoidance_layers()

A bitfield determining the avoidance layers for this NavigationAgent. Other agents with a matching bit on the avoidance_mask will avoid this agent.

int avoidance_mask = 1 🔗

void set_avoidance_mask(value: int)
int get_avoidance_mask()

A bitfield determining what other avoidance agents and obstacles this NavigationAgent will avoid when a bit matches at least one of their avoidance_layers.

float avoidance_priority = 1.0 🔗

void set_avoidance_priority(value: float)
float get_avoidance_priority()

The agent does not adjust the velocity for other agents that would match the avoidance_mask but have a lower avoidance_priority. This in turn makes the other agents with lower priority adjust their velocities even more to avoid collision with this agent.

bool debug_enabled = false 🔗

void set_debug_enabled(value: bool)
bool get_debug_enabled()

If true shows debug visuals for this agent.

Color debug_path_custom_color = Color(1, 1, 1, 1) 🔗

void set_debug_path_custom_color(value: Color)
Color get_debug_path_custom_color()

If debug_use_custom is true uses this color for this agent instead of global color.

float debug_path_custom_point_size = 4.0 🔗

void set_debug_path_custom_point_size(value: float)
float get_debug_path_custom_point_size()

If debug_use_custom is true uses this rasterized point size for rendering path points for this agent instead of global point size.

bool debug_use_custom = false 🔗

void set_debug_use_custom(value: bool)
bool get_debug_use_custom()

If true uses the defined debug_path_custom_color for this agent instead of global color.

float height = 1.0 🔗

void set_height(value: float)
float get_height()

The height of the avoidance agent. Agents will ignore other agents or obstacles that are above or below their current position + height in 2D avoidance. Does nothing in 3D avoidance which uses radius spheres alone.

bool keep_y_velocity = true 🔗

void set_keep_y_velocity(value: bool)
bool get_keep_y_velocity()

If true, and the agent uses 2D avoidance, it will remember the set y-axis velocity and reapply it after the avoidance step. While 2D avoidance has no y-axis and simulates on a flat plane this setting can help to soften the most obvious clipping on uneven 3D geometry.

int max_neighbors = 10 🔗

void set_max_neighbors(value: int)
int get_max_neighbors()

The maximum number of neighbors for the agent to consider.

float max_speed = 10.0 🔗

void set_max_speed(value: float)
float get_max_speed()

The maximum speed that an agent can move.

int navigation_layers = 1 🔗

void set_navigation_layers(value: int)
int get_navigation_layers()

A bitfield determining which navigation layers of navigation regions this agent will use to calculate a path. Changing it during runtime will clear the current navigation path and generate a new one, according to the new navigation layers.

float neighbor_distance = 50.0 🔗

void set_neighbor_distance(value: float)
float get_neighbor_distance()

The distance to search for other agents.

float path_desired_distance = 1.0 🔗

void set_path_desired_distance(value: float)
float get_path_desired_distance()

The distance threshold before a path point is considered to be reached. This allows agents to not have to hit a path point on the path exactly, but only to reach its general area. If this value is set too high, the NavigationAgent will skip points on the path, which can lead to it leaving the navigation mesh. If this value is set too low, the NavigationAgent will be stuck in a repath loop because it will constantly overshoot the distance to the next point on each physics frame update.

float path_height_offset = 0.0 🔗

void set_path_height_offset(value: float)
float get_path_height_offset()

The height offset is subtracted from the y-axis value of any vector path position for this NavigationAgent. The NavigationAgent height offset does not change or influence the navigation mesh or pathfinding query result. Additional navigation maps that use regions with navigation meshes that the developer baked with appropriate agent radius or height values are required to support different-sized agents.

float path_max_distance = 5.0 🔗

void set_path_max_distance(value: float)
float get_path_max_distance()

The maximum distance the agent is allowed away from the ideal path to the final position. This can happen due to trying to avoid collisions. When the maximum distance is exceeded, it recalculates the ideal path.

BitField[PathMetadataFlags] path_metadata_flags = 7 🔗

void set_path_metadata_flags(value: BitField[PathMetadataFlags])
BitField[PathMetadataFlags] get_path_metadata_flags()

Additional information to return with the navigation path.

PathPostProcessing path_postprocessing = 0 🔗

void set_path_postprocessing(value: PathPostProcessing)
PathPostProcessing get_path_postprocessing()

The path postprocessing applied to the raw path corridor found by the pathfinding_algorithm.

PathfindingAlgorithm pathfinding_algorithm = 0 🔗

void set_pathfinding_algorithm(value: PathfindingAlgorithm)
PathfindingAlgorithm get_pathfinding_algorithm()

The pathfinding algorithm used in the path query.

float radius = 0.5 🔗

void set_radius(value: float)
float get_radius()

The radius of the avoidance agent. This is the "body" of the avoidance agent and not the avoidance maneuver starting radius (which is controlled by neighbor_distance).

Does not affect normal pathfinding. To change an actor's pathfinding radius bake NavigationMesh resources with a different NavigationMesh.agent_radius property and use different navigation maps for each actor size.

float simplify_epsilon = 0.0 🔗

void set_simplify_epsilon(value: float)
float get_simplify_epsilon()

The path simplification amount in worlds units.

bool simplify_path = false 🔗

void set_simplify_path(value: bool)
bool get_simplify_path()

If true a simplified version of the path will be returned with less critical path points removed. The simplification amount is controlled by simplify_epsilon. The simplification uses a variant of Ramer-Douglas-Peucker algorithm for curve point decimation.

Path simplification can be helpful to mitigate various path following issues that can arise with certain agent types and script behaviors. E.g. "steering" agents or avoidance in "open fields".

float target_desired_distance = 1.0 🔗

void set_target_desired_distance(value: float)
float get_target_desired_distance()

The distance threshold before the target is considered to be reached. On reaching the target, target_reached is emitted and navigation ends (see is_navigation_finished and navigation_finished).

You can make navigation end early by setting this property to a value greater than path_desired_distance (navigation will end before reaching the last waypoint).

You can also make navigation end closer to the target than each individual path position by setting this property to a value lower than path_desired_distance (navigation won't immediately end when reaching the last waypoint). However, if the value set is too low, the agent will be stuck in a repath loop because it will constantly overshoot the distance to the target on each physics frame update.

Vector3 target_position = Vector3(0, 0, 0) 🔗

void set_target_position(value: Vector3)
Vector3 get_target_position()

If set, a new navigation path from the current agent position to the target_position is requested from the NavigationServer.

float time_horizon_agents = 1.0 🔗

void set_time_horizon_agents(value: float)
float get_time_horizon_agents()

The minimal amount of time for which this agent's velocities, that are computed with the collision avoidance algorithm, are safe with respect to other agents. The larger the number, the sooner the agent will respond to other agents, but less freedom in choosing its velocities. A too high value will slow down agents movement considerably. Must be positive.

float time_horizon_obstacles = 0.0 🔗

void set_time_horizon_obstacles(value: float)
float get_time_horizon_obstacles()

The minimal amount of time for which this agent's velocities, that are computed with the collision avoidance algorithm, are safe with respect to static avoidance obstacles. The larger the number, the sooner the agent will respond to static avoidance obstacles, but less freedom in choosing its velocities. A too high value will slow down agents movement considerably. Must be positive.

bool use_3d_avoidance = false 🔗

void set_use_3d_avoidance(value: bool)
bool get_use_3d_avoidance()

If true, the agent calculates avoidance velocities in 3D omnidirectionally, e.g. for games that take place in air, underwater or space. Agents using 3D avoidance only avoid other agents using 3D avoidance, and react to radius-based avoidance obstacles. They ignore any vertex-based obstacles.

If false, the agent calculates avoidance velocities in 2D along the x and z-axes, ignoring the y-axis. Agents using 2D avoidance only avoid other agents using 2D avoidance, and react to radius-based avoidance obstacles or vertex-based avoidance obstacles. Other agents using 2D avoidance that are below or above their current position including height are ignored.

Vector3 velocity = Vector3(0, 0, 0) 🔗

void set_velocity(value: Vector3)
Vector3 get_velocity()

Sets the new wanted velocity for the agent. The avoidance simulation will try to fulfill this velocity if possible but will modify it to avoid collision with other agents and obstacles. When an agent is teleported to a new position, use set_velocity_forced as well to reset the internal simulation velocity.

Method Descriptions

float distance_to_target() const 🔗

Returns the distance to the target position, using the agent's global position. The user must set target_position in order for this to be accurate.

bool get_avoidance_layer_value(layer_number: int) const 🔗

Returns whether or not the specified layer of the avoidance_layers bitmask is enabled, given a layer_number between 1 and 32.

bool get_avoidance_mask_value(mask_number: int) const 🔗

Returns whether or not the specified mask of the avoidance_mask bitmask is enabled, given a mask_number between 1 and 32.

Returns this agent's current path from start to finish in global coordinates. The path only updates when the target position is changed or the agent requires a repath. The path array is not intended to be used in direct path movement as the agent has its own internal path logic that would get corrupted by changing the path array manually. Use the intended get_next_path_position once every physics frame to receive the next path point for the agents movement as this function also updates the internal path logic.

Returns which index the agent is currently on in the navigation path's PackedVector3Array.

Returns the path query result for the path the agent is currently following.

Vector3 get_final_position() 🔗

Returns the reachable final position of the current navigation path in global coordinates. This position can change if the agent needs to update the navigation path which makes the agent emit the path_changed signal.

Returns whether or not the specified layer of the navigation_layers bitmask is enabled, given a layer_number between 1 and 32.

Returns the RID of the navigation map for this NavigationAgent node. This function returns always the map set on the NavigationAgent node and not the map of the abstract agent on the NavigationServer. If the agent map is changed directly with the NavigationServer API the NavigationAgent node will not be aware of the map change. Use set_navigation_map to change the navigation map for the NavigationAgent and also update the agent on the NavigationServer.

Vector3 get_next_path_position() 🔗

Returns the next position in global coordinates that can be moved to, making sure that there are no static objects in the way. If the agent does not have a navigation path, it will return the position of the agent's parent. The use of this function once every physics frame is required to update the internal path logic of the NavigationAgent.

RID get_rid() const 🔗

Returns the RID of this agent on the NavigationServer3D.

Returns true if the agent's navigation has finished. If the target is reachable, navigation ends when the target is reached. If the target is unreachable, navigation ends when the last waypoint of the path is reached.

Note: While true prefer to stop calling update functions like get_next_path_position. This avoids jittering the standing agent due to calling repeated path updates.

bool is_target_reachable() 🔗

Returns true if get_final_position is within target_desired_distance of the target_position.

bool is_target_reached() const 🔗

Returns true if the agent reached the target, i.e. the agent moved within target_desired_distance of the target_position. It may not always be possible to reach the target but it should always be possible to reach the final position. See get_final_position.

void set_avoidance_layer_value(layer_number: int, value: bool) 🔗

Based on value, enables or disables the specified layer in the avoidance_layers bitmask, given a layer_number between 1 and 32.

void set_avoidance_mask_value(mask_number: int, value: bool) 🔗

Based on value, enables or disables the specified mask in the avoidance_mask bitmask, given a mask_number between 1 and 32.

Based on value, enables or disables the specified layer in the navigation_layers bitmask, given a layer_number between 1 and 32.

Sets the RID of the navigation map this NavigationAgent node should use and also updates the agent on the NavigationServer.

void set_velocity_forced(velocity: Vector3) 🔗

Replaces the internal velocity in the collision avoidance simulation with velocity. When an agent is teleported to a new position this function should be used in the same frame. If called frequently this function can get agents stuck.