`fastsim.planner`

8 个类

Abstract planner interface for motion planning backends. Implement Planner to integrate a custom IK/FK/motion-plan solver. CuroboPlanner provides a ready-to-use GPU-accelerated implementation.

Planner CuroboPlanner OmplPlanner ArmPlanner NavigationPlanner AStarDWAPlanner CuroboMpcNavigationPlanner PlannerFactory

class

Planner

抽象类7 个方法

fastsim.planner.planner.Planner继承自ABC

Abstract base class for all planners (arm and navigation). Provides minimal shared interface.

Planner defines the shared interface for both ArmPlanner (manipulator planning) and NavigationPlanner (mobile base planning). Subclasses implement update_world() and get_world_representation(). For arm-specific methods (solve_ik, gen_motion_plan, etc.), see ArmPlanner. For navigation methods (plan_path, compute_velocity, etc.), see NavigationPlanner.

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: list, is_goal_set: bool = False, retract_joint_position: list | None = None) → dict

Solve inverse kinematics for a target EE pose.

参数

参数名	类型	默认值	说明
`base_frame_ee_pose`	list	—	Target EE pose [x, y, z, qw, qx, qy, qz].
`is_goal_set`opt	bool	`False`	Whether a motion goal has already been set in the planner state.
`retract_joint_position`opt	list\|None	`None`	Optional retract seed for singularity avoidance.

返回值

dict — Joint position solution dict.

solve_fkmethod

def solve_fk(self, joint_positions: list, joint_names: list) → dict

Compute forward kinematics for a joint configuration.

参数

参数名	类型	默认值	说明
`joint_positions`	list	—	Joint position values.
`joint_names`	list	—	Corresponding joint names.

返回值

dict — End-effector pose dict.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: list, is_goal_set: bool = False) → dict

Generate a collision-free joint trajectory.

参数

参数名	类型	默认值	说明
`start_state`	dict	—	Starting joint state dict.
`base_frame_ee_pose`	list	—	Goal EE pose.
`is_goal_set`opt	bool	`False`	Whether the planner goal is pre-set.

返回值

dict — Planned trajectory dict.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list, link_name: str, attach_offset: list) → None

参数

参数名	类型	默认值	说明
`joint_state`	dict	—	Current joint state at time of attachment.
`object_prim_paths`	list	—	USD prim paths of objects to attach.
`link_name`	str	—	Robot link name for the attachment.
`attach_offset`	list	—	Pose offset [x, y, z, qw, qx, qy, qz] from the link.

返回值

None

detach_objectsmethod

def detach_objects(self, link_name: str) → None

Remove attached objects from the planner collision model.

参数

参数名	类型	默认值	说明
`link_name`	str	—	The link name from which to detach.

返回值

None

get_plannable_joint_namesmethod

def get_plannable_joint_names(self) → list

Return the list of joints this planner controls.

返回值

list — List of plannable joint name strings.

get_locked_jointmethod

def get_locked_joint(self) → list

Return joints excluded from planning (locked).

返回值

list — List of locked joint name strings.

class

CuroboPlanner

2 个方法

fastsim.planner.curobo_planner.CuroboPlanner继承自ArmPlanner

GPU-accelerated arm motion planner backed by the cuRobo library.

CuroboPlanner implements all Planner abstract methods using cuRobo's batched IK, FK, and motion-generation solvers. Requires a CUDA-capable GPU.

get_world_representationmethod

def get_world_representation(self) → Any

Return the cuRobo world collision model (voxel or mesh).

返回值

Any — cuRobo world representation object.

get_robot_representationmethod

def get_robot_representation(self) → Any

Return the cuRobo robot self-collision model.

返回值

Any — cuRobo robot representation object.

继承自ArmPlanner4

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: Any) → dict

Solve inverse kinematics for one or more target EE poses.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: Any) → dict

Generate a collision-free trajectory to the target EE pose.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list) → bool

Attach objects to the robot model in the planner world.

detach_objectsmethod

def detach_objects(self)

Detach previously attached objects from the planner world.

class

OmplPlanner

0 个方法

fastsim.planner.ompl_planner.OmplPlanner继承自ArmPlanner

CPU-based arm motion planner using the Open Motion Planning Library (OMPL) with PyBullet collision checking.

OmplPlanner reads collision geometry from SceneManager and uses PyBullet for collision detection. Does not require a GPU. Works with all simulator backends.

继承自ArmPlanner4

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: Any) → dict

Solve inverse kinematics for one or more target EE poses.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: Any) → dict

Generate a collision-free trajectory to the target EE pose.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list) → bool

Attach objects to the robot model in the planner world.

detach_objectsmethod

def detach_objects(self)

Detach previously attached objects from the planner world.

class

ArmPlanner

抽象类4 个方法

fastsim.planner.arm_planner.ArmPlanner继承自Planner

Abstract base for manipulator arm planners (IK, FK, motion planning, attach/detach).

ArmPlanner extends Planner with arm-specific abstract methods. CuroboPlanner and OmplPlanner both inherit from this class.

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: Any) → dict

Solve inverse kinematics for one or more target EE poses.

参数

参数名	类型	默认值	说明
`base_frame_ee_pose`	Any	—	Target EE pose(s) in base frame.

返回值

dict — IK solution dict with joint positions.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: Any) → dict

Generate a collision-free trajectory to the target EE pose.

参数

参数名	类型	默认值	说明
`start_state`	dict	—	Current joint state.
`base_frame_ee_pose`	Any	—	Target EE pose(s).

返回值

dict — Motion plan with joint trajectory.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list) → bool

Attach objects to the robot model in the planner world.

参数

参数名	类型	默认值	说明
`joint_state`	dict	—	Current joint state.
`object_prim_paths`	list	—	Prim paths of objects to attach.

返回值

bool — True if attachment succeeded.

detach_objectsmethod

def detach_objects(self)

Detach previously attached objects from the planner world.

继承自Planner7

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: list, is_goal_set: bool = False, retract_joint_position: list | None = None) → dict

Solve inverse kinematics for a target EE pose.

solve_fkmethod

def solve_fk(self, joint_positions: list, joint_names: list) → dict

Compute forward kinematics for a joint configuration.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: list, is_goal_set: bool = False) → dict

Generate a collision-free joint trajectory.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list, link_name: str, attach_offset: list) → None

detach_objectsmethod

def detach_objects(self, link_name: str) → None

Remove attached objects from the planner collision model.

get_plannable_joint_namesmethod

def get_plannable_joint_names(self) → list

Return the list of joints this planner controls.

get_locked_jointmethod

def get_locked_joint(self) → list

Return joints excluded from planning (locked).

class

NavigationPlanner

抽象类5 个方法

fastsim.planner.navigation_planner.NavigationPlanner继承自Planner

Abstract base for mobile base navigation planners (path planning, velocity control).

NavigationPlanner extends Planner with navigation-specific methods: plan_path, compute_velocity, set_goal, cancel, is_goal_reached. The A*+DWA planner is the built-in implementation.

plan_pathmethod

def plan_path(self, start: Any, goal: Any) → Any

Plan a path from start to goal on the 2D ground plane.

参数

参数名	类型	默认值	说明
`start`	Any	—	Start pose (Pose2D: x, y, theta).
`goal`	Any	—	Goal pose (Pose2D: x, y, theta).

返回值

Any — NavigationResult with success flag and path.

compute_velocitymethod

def compute_velocity(self, current_pose: Any, current_velocity: Any) → Any

Compute the next velocity command (DWA local planning).

参数

参数名	类型	默认值	说明
`current_pose`	Any	—	Current Pose2D.
`current_velocity`	Any	—	Current Twist2D.

返回值

Any — Twist2D velocity command.

set_goalmethod

def set_goal(self, goal: Any) → bool

Set the navigation goal and plan a path.

参数

参数名	类型	默认值	说明
`goal`	Any	—	Goal Pose2D.

返回值

bool — True if path was found.

is_goal_reachedmethod

def is_goal_reached(self, pos_tolerance: float = 0.05, angle_tolerance: float = 0.1) → bool

Check if the robot has reached the goal within tolerance.

参数

参数名	类型	默认值	说明
`pos_tolerance`opt	float	`0.05`
`angle_tolerance`opt	float	`0.1`

返回值

bool — True if goal reached.

cancelmethod

def cancel(self)

Cancel the current navigation goal.

继承自Planner7

solve_ikmethod

def solve_ik(self, base_frame_ee_pose: list, is_goal_set: bool = False, retract_joint_position: list | None = None) → dict

Solve inverse kinematics for a target EE pose.

solve_fkmethod

def solve_fk(self, joint_positions: list, joint_names: list) → dict

Compute forward kinematics for a joint configuration.

gen_motion_planmethod

def gen_motion_plan(self, start_state: dict, base_frame_ee_pose: list, is_goal_set: bool = False) → dict

Generate a collision-free joint trajectory.

attach_objectsmethod

def attach_objects(self, joint_state: dict, object_prim_paths: list, link_name: str, attach_offset: list) → None

detach_objectsmethod

def detach_objects(self, link_name: str) → None

Remove attached objects from the planner collision model.

get_plannable_joint_namesmethod

def get_plannable_joint_names(self) → list

Return the list of joints this planner controls.

get_locked_jointmethod

def get_locked_joint(self) → list

Return joints excluded from planning (locked).

class

AStarDWAPlanner

0 个方法

fastsim.planner.planners.astar_dwa.astar_dwa_planner.AStarDWAPlanner继承自NavigationPlanner

A*+DWA hybrid navigation planner for mobile bases.

Combines A* global path planning on an occupancy grid with Dynamic Window Approach (DWA) local velocity optimization. Builds occupancy grid from scene objects or sensor data. Config: AStarDWAPlannerConfig now extends NavigationPlannerConfig.

继承自NavigationPlanner5

plan_pathmethod

def plan_path(self, start: Any, goal: Any) → Any

Plan a path from start to goal on the 2D ground plane.

compute_velocitymethod

def compute_velocity(self, current_pose: Any, current_velocity: Any) → Any

Compute the next velocity command (DWA local planning).

set_goalmethod

def set_goal(self, goal: Any) → bool

Set the navigation goal and plan a path.

is_goal_reachedmethod

def is_goal_reached(self, pos_tolerance: float = 0.05, angle_tolerance: float = 0.1) → bool

Check if the robot has reached the goal within tolerance.

cancelmethod

def cancel(self)

Cancel the current navigation goal.

class

CuroboMpcNavigationPlanner

0 个方法

fastsim.planner.planners.curobo_mpc_nav.curobo_mpc_nav_planner.CuroboMpcNavigationPlanner继承自NavigationPlanner

GPU MPC navigation planner. Walks an executable_path of rotate / translate sub-poses; uses curobo MpcSolver per tick.

Refreshes both A*-based occupancy grid and curobo world on update_world. Lazy-initializes MpcSolver / WorldConfig / robot handle on first use. Writes joint position targets (3 base joints) directly via compute_velocity instead of returning a Twist2D. Stereotype: curobo_mpc_nav.

继承自NavigationPlanner5

plan_pathmethod

def plan_path(self, start: Any, goal: Any) → Any

Plan a path from start to goal on the 2D ground plane.

compute_velocitymethod

def compute_velocity(self, current_pose: Any, current_velocity: Any) → Any

Compute the next velocity command (DWA local planning).

set_goalmethod

def set_goal(self, goal: Any) → bool

Set the navigation goal and plan a path.

is_goal_reachedmethod

def is_goal_reached(self, pos_tolerance: float = 0.05, angle_tolerance: float = 0.1) → bool

Check if the robot has reached the goal within tolerance.

cancelmethod

def cancel(self)

Cancel the current navigation goal.

class

PlannerFactory

静态类1 个方法

fastsim.planner.planner_factory.PlannerFactory

Static factory for creating Planner instances from a config dict.

createstaticmethod

@staticmethod def create(robot_name: str, config: dict) → Planner

Instantiate a Planner for a given robot name and config.

参数

参数名	类型	默认值	说明
`robot_name`	str	—	Name of the robot this planner will control.
`config`	dict	—	Planner configuration dict specifying type and parameters.

返回值

Planner — Constructed Planner instance.