class mola::Kitti360Dataset

Overview

Each “sequence” directory contains these sensor streams:

image_0 & image_1 : Perspective cameras
image_2 & image_3 : Fish-eye cameras (TO-DO)
lidar : Velodyne 3D LIDAR. Published as mrpt::obs::CObservationPointCloud with X,Y,Z,I,T channels. The timestamp (T) channel is estimated from azimuth of points (if generate_lidar_timestamps param is true), with generated timestamps between [-0.05,+0.05] seconds.
Ground truth poses

The sequence to load is determined by the sequence parameter (e.g. via config yaml file), from these possible values:

sequence: 00|02|03|04|05|06|07|08|09|10|18|test_0|test_1|test_2|test_3

Expected contents under base_dir directory (the result of using the download scripts provided by the KITTI-360 web):

KITTI-360/
├── calibration
│  ├── calib_cam_to_pose.txt
│  ├── calib_cam_to_velo.txt
│  ├── calib_sick_to_velo.txt
│  ├── image_02.yaml
│  ├── image_03.yaml
│  └── perspective.txt
│
├── data_3d_raw
│   ├ 2013_05_28_drive_00xx_sync
│   │ ├── cam0_to_world.txt
│   │ ├── poses.txt
│   │ └── velodyne_points
│   │     ├── timestamps.txt
│   │     └── data
│   │         ├── 0000000000.bin
│   │         ├── 0000000001.bin
... ...           ...
├── data_2d_raw
│   ├── 2013_05_28_drive_00xx_sync
│   │   ├── image_00
│   │   │   ├── data_rect
│   │   │   │   ├ 0000000000.png
│   │   │   │   ├ 0000000001.png
│   │   │   │   ...
│   │   │   └── timestamps.txt
│   │   └── image_01
│   │       ├── data_rect
│   │       │   ├ 0000000000.png
│   │       │   ├ 0000000001.png
│   │       │   ...
│   │       └── timestamps.txt
... ...
├── data_poses/
│   ├── 2013_05_28_drive_00xx_sync
│   │   ├── cam0_to_world.txt
│   │   └── poses.txt
... ...
└──  data_3d_test_slam/
    ├── test_0
    │   └── 2013_05_28_drive_0008_sync
    │       └── velodyne_points
    │           ├── data
    │           └── timestamps.txt
    ├── test_1
    │   └── 2013_05_28_drive_0008_sync
    │       └── velodyne_points
    │           ├── data
    │           └── timestamps.txt
    ├── test_2
    │   └── 2013_05_28_drive_0004_sync
    │       └── velodyne_points
    │           ├── data
    │           └── timestamps.txt
    └── test_3
        └── 2013_05_28_drive_0002_sync
            └── velodyne_points
                ├── data
                └── timestamps.txt

Example base_dir : /mnt/storage/KITTI-360/ (normally read from environment variable KITTI360_DATASET in mola-cli launch files).
Sequences: 01, 02, etc.

#include <Kitti360Dataset.h>

class Kitti360Dataset:
    public mola::RawDataSourceBase,
    public mola::OfflineDatasetSource,
    public mola::Dataset_UI
{
public:
    // fields

    double VERTICAL_ANGLE_OFFSET = mrpt::DEG2RAD(0.205);

    // methods

    virtual void spinOnce();
    virtual bool hasGroundTruthTrajectory() const;
    virtual trajectory_t getGroundTruthTrajectory() const;
    std::shared_ptr<mrpt::obs::CObservation> getPointCloud(timestep_t step) const;

    std::shared_ptr<mrpt::obs::CObservationImage> getImage(
        const unsigned int cam_idx,
        timestep_t step
        ) const;

    virtual size_t datasetSize() const;
    virtual mrpt::obs::CSensoryFrame::Ptr datasetGetObservations(size_t timestep) const;
    virtual size_t datasetUI_size() const;
    virtual size_t datasetUI_lastQueriedTimestep() const;
    virtual double datasetUI_playback_speed() const;
    virtual void datasetUI_playback_speed(double speed);
    virtual bool datasetUI_paused() const;
    virtual void datasetUI_paused(bool paused);
    virtual void datasetUI_teleport(size_t timestep);
};

Inherited Members

public:
    // structs

    struct DiagnosticsOutput;
    struct SensorViewerImpl;

    // methods

    virtual void initialize(const Yaml& cfg) = 0;
    virtual void spinOnce() = 0;
    ExecutableBase& operator = (const ExecutableBase&);
    ExecutableBase& operator = (ExecutableBase&&);
    RawDataSourceBase& operator = (const RawDataSourceBase&);
    RawDataSourceBase& operator = (RawDataSourceBase&&);
    virtual size_t datasetSize() const = 0;
    virtual mrpt::obs::CSensoryFrame::Ptr datasetGetObservations(size_t timestep) const = 0;
    virtual bool hasGroundTruthTrajectory() const;
    virtual size_t datasetUI_size() const = 0;
    virtual size_t datasetUI_lastQueriedTimestep() const = 0;
    virtual double datasetUI_playback_speed() const = 0;
    virtual void datasetUI_playback_speed(double speed) = 0;
    virtual bool datasetUI_paused() const = 0;
    virtual void datasetUI_paused(bool paused) = 0;
    virtual void datasetUI_teleport(size_t timestep) = 0;
    virtual std::optional<double> datasetUI_time() const;
    virtual std::optional<double> datasetUI_total_time() const;

Fields

double VERTICAL_ANGLE_OFFSET = mrpt::DEG2RAD(0.205)

See: “IMLS-SLAM: scan-to-model matching based on 3D data”, JE Deschaud, 2018.

Methods

virtual void spinOnce()

Runs any required action on a timely manner

virtual bool hasGroundTruthTrajectory() const

Returns true if a groundtruth is available for the vehicle trajectory.