2D/3D points and poses

Lie Algebra methods for SO(2),SO(3),SE(2),SE(3)

// namespaces

namespace mrpt::poses;
    namespace mrpt::poses::detail;
    namespace mrpt::poses::internal;
    namespace mrpt::poses::Lie;

// enums

enum mrpt::poses::TInterpolatorMethod;

// classes

template <class DERIVEDCLASS, std::size_t DIM>
class mrpt::poses::CPoint;

class mrpt::poses::CPoint2D;
class mrpt::poses::CPoint3D;

template <class DERIVEDCLASS, std::size_t DIM>
class mrpt::poses::CPose;

class mrpt::poses::CPose2D;
class mrpt::poses::CPose3D;
class mrpt::poses::CPose3DQuat;

template <class DERIVEDCLASS, std::size_t DIM>
class mrpt::poses::CPoseOrPoint;

class mrpt::poses::CPoses2DSequence;
class mrpt::poses::CPoses3DSequence;
class mrpt::poses::CRobot2DPoseEstimator;

template <int DIM>
class mrpt::poses::FrameTransformer;

template <int DIM>
class mrpt::poses::FrameTransformerInterface;

template <>
class mrpt::poses::SE_average<3>;

template <>
class mrpt::poses::SE_average<2>;

template <size_t DOF>
class mrpt::poses::SE_average;

template <>
class mrpt::poses::SO_average<2>;

template <size_t DOF>
class mrpt::poses::SO_average;

template <>
class mrpt::poses::SO_average<3>;

// global functions

SensorToPoseMap mrpt::poses::sensor_poses_from_yaml(const mrpt::containers::yaml& d, const std::string& referenceFrame = "base_link");
SensorToPoseMap mrpt::poses::sensor_poses_from_yaml_file(const std::string& filename, const std::string& referenceFrame = "base_link");

Global Functions

SensorToPoseMap mrpt::poses::sensor_poses_from_yaml(
    const mrpt::containers::yaml& d,
    const std::string& referenceFrame = "base_link"
    )

Alternative to sensor_poses_from_yaml_file() where the yaml map inside sensors: ... is directly passed programatically.

SensorToPoseMap mrpt::poses::sensor_poses_from_yaml_file(
    const std::string& filename,
    const std::string& referenceFrame = "base_link"
    )

Utility to parse a YAML file with the extrinsic calibration of sensors.

Each YAML map entry defines a sensorLabel, and for each one an extrinsics map containing the SE(3) relative pose between the parent frame and this sensor. The pose is given as a quaternion and a translation.

The expected file contents is like:

*  # My YAML file:
*  sensors:  # Note: sensor_poses_from_yaml() expects **this node** as input
*    camera:
*      extrinsics:
*        quaternion: [qx, qy, qz, qw]
*        translation: [tx, ty, tz]
*    parent: base_link
*    imu:
*      extrinsics:
*        quaternion: [qx, qy, qz, qw]
*        translation: [tx, ty, tz]
*    parent: camera
*

Following the common ROS conventions, the robot reference frame is assumed to be base_link (default).

Of course, this mechanism of defining a tree of sensor poses in a YAML file only works for static (rigid) sensor assemblies, where the transformations between them is always static.

The data is returned as a std::map from sensor labels to poses within the robot reference frame.

This function takes as input the YAML filename to load.