US 12,394,216 B2
Method for predicting an ego-lane for a vehicle
Alexander Lengsfeld, Bad Muender (DE); and Philip Lenz, Holle (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Aug. 9, 2022, as Appl. No. 17/883,846.
Claims priority of application No. 10 2021 208 830.9 (DE), filed on Aug. 12, 2021.
Prior Publication US 2023/0052594 A1, Feb. 16, 2023
Int. Cl. G06V 20/56 (2022.01); B60W 30/10 (2006.01); G01C 21/00 (2006.01); G06V 10/82 (2022.01)
CPC G06V 20/588 (2022.01) [G01C 21/3819 (2020.08); G06V 10/82 (2022.01); B60W 30/10 (2013.01); B60W 2420/403 (2013.01); B60W 2556/40 (2020.02); G06V 2201/12 (2022.01)] 13 Claims
OG exemplary drawing
 
1. A method for predicting an ego-lane for a vehicle, comprising the following steps:
receiving at least one image captured by at last one camera sensor of the vehicle an in which a lane is depicted;
providing the received at least one image as input into a trained neural network that is trained via regression to directly convert the input at least one image into output of a representation coordinates of a center line that extends in a center of the lane without identifying boundaries of the lane, wherein the output is in a form of a plurality of parameters;
generating the center line based on the output parameters of the center line;
identifying the generated center line of the lane as at least part of the predicted ego-lane of the vehicle; and
providing the predicted ego-lane.