LiDAR data conversion apparatus and method for training various types of autonomous vehicles by using pre-acquired data

1 . A processor-implemented method for converting LiDAR data, the method comprising:

receiving an input of first LiDAR data which is pre-acquired through a first LiDAR sensor mounted in a first vehicle;

converting the inputted first LiDAR data into second LiDAR data which is acquired through and corresponding to a second LiDAR sensor mounted in a second vehicle having a different installation height and a different mounting position from the first LiDAR sensor mounted in the first vehicle; and

outputting the converted second LiDAR data to be used as a training dataset for the second vehicle in training autonomous driving features for the second vehicle,

wherein the converting comprises converting the first LiDAR data into LiDAR data on a reference coordinate system, adjusting the LiDAR data based on a respective difference between the respective installation heights and the respective mounting positions of the first LiDAR sensor and the second LiDAR sensor, and converting the adjusted LiDAR data into the second LiDAR data which is LiDAR data on a coordinate system of the second LiDAR sensor, such that learning performance by pre-acquired sensor data is improved and utilization of the pre-acquired sensor data is increased.

2 . The method of claim 1 , wherein the reference coordinate system has a center point of a rear wheel axis of a vehicle as an original point, and

wherein the reference coordinate system has an X axis facing in a forward direction from the original point, a Y axis facing in a side direction, and a Z axis facing in a direction from the ground toward a space.

3 . The method of claim 1 , further comprising performing spherical projection with respect to the second LiDAR data with reference to the coordinate axes of the second LIDAR sensor.

4 . The method of claim 3 , wherein the performing the spherical projection comprises, when two or more pieces of data overlap at a same position, leaving data that has a smaller range value, and discarding data that has a larger range value.

5 . The method of claim 3 , further comprising performing interpolation with respect to the second LiDAR data which undergoes the spherical projection.

6 . The method of claim 5 , further comprising performing semantic segmentation with respect to the inputted first LiDAR data.

7 . The method of claim 6 , wherein the performing the interpolation comprises performing interpolation based on a result of the semantic segmentation.

8 . The method of claim 7 , wherein the performing the interpolation comprises performing interpolation with reference to left and right data after performing interpolation with reference to upper and lower data.

9 . The method of claim 5 , further comprising converting the interpolated second LiDAR data into cloud point data,

wherein the outputting comprises outputting the second LiDAR data converted into the cloud point data.

10 . A LIDAR data conversion apparatus comprising:

one or more processors configured to:

receive an input of first LiDAR data which is pre-acquired through a first LiDAR sensor mounted in a first vehicle;

convert the inputted first LiDAR data into second LiDAR data which is acquired through and corresponding to a second LiDAR sensor mounted in a second vehicle having a different installation height and a different mounting position from the first LiDAR sensor mounted in the first vehicle; and

output the converted second LiDAR data to be used as a training dataset for the second vehicle in training autonomous driving features for the second vehicle,

wherein the one or more processors are configured to convert the first LiDAR data into LIDAR data on a reference coordinate system, adjusting the LiDAR data based on a respective difference between the respective installation heights and the respective mounting positions of the first LiDAR sensor and the second LiDAR sensor, and convert the converted LiDAR data into the second LiDAR data which is LiDAR data on a coordinate system of the second LiDAR sensor, such that learning performance by pre-acquired sensor data is improved and utilization of the pre-acquired sensor data is increased.

11 . The apparatus of claim 10 , wherein the reference coordinate system has a center point of a rear wheel axis of a vehicle as an original point, and

wherein the reference coordinate system has an X axis facing in a forward direction from the original point, a Y axis facing in a side direction, and a Z axis facing in a direction from the ground toward a space.

12 . The apparatus of claim 10 , further comprising performing spherical projection with respect to the second LiDAR data with reference to the coordinate axes of the second LiDAR sensor.

13 . The apparatus of claim 12 , wherein, for the performing the spherical projection, the one or more processors are configured to, when two or more pieces of data overlap at a same position, leave data that has a smaller range value, and discard data that has a larger range value.

14 . The apparatus of claim 12 , wherein the one or more processors are configured to perform interpolation with respect to the second LiDAR data which undergoes the spherical projection.

15 . The apparatus of claim 14 , wherein the one or more processors are configured to perform semantic segmentation with respect to the inputted first LiDAR data.

16 . The apparatus of claim 15 , wherein for the performing the interpolation, the one or more processors are configured to perform interpolation based on a result of the semantic segmentation.

17 . The apparatus of claim 16 , wherein, for the performing the interpolation, the one or more processors are configured to perform interpolation with reference to left and right data after performing interpolation with reference to upper and lower data.

18 . The apparatus of claim 14 , wherein the one or more processors are configured to convert the interpolated second LiDAR data into cloud point data, and output the second LiDAR data converted into the cloud point data.

19 . A method for converting LiDAR data, the method comprising:

receiving an input of first LiDAR data which is pre-acquired through a first LiDAR sensor mounted in a first vehicle;

converting the inputted first LiDAR data into second LiDAR data which is acquired through and corresponding to a second LiDAR sensor mounted in a second vehicle having a different installation height and a different mounting position from the first LiDAR sensor mounted in the first vehicle; and

outputting the converted second LiDAR data to be used as a training dataset for the second vehicle in training autonomous driving features for the second vehicle,

wherein the first LiDAR sensor and the second LiDAR sensor have different installation heights, and

wherein the converting comprises converting the first LiDAR data into LiDAR data on a reference coordinate system, adjusting the LiDAR data based on a respective difference between the respective installation heights and the respective mounting positions of the first LIDAR sensor and the second LiDAR sensor, and converting the adjusted LiDAR data into the second LiDAR data which is LiDAR data on a coordinate system of the second LiDAR sensor, such that learning performance by pre-acquired sensor data is improved and utilization of the pre-acquired sensor data is increased.