Device and method for improving assistance systems for lateral vehicle movements
View Patent ↗A system includes at least one electronic control unit, which performs a method including consecutively or simultaneously, determining a future turning maneuver of the ego vehicle, detecting information relating to the lane markings and the number of available lanes in the environment in front of and next to the ego vehicle, and determining whether at least one further road user is in a relevant lane next to or behind the lane of the ego vehicle. If so, a future intended movement of the road user is determined from the information relating to the lane markings and the number of available lanes. And if it is determined that the ego-vehicle and the road user are on an at least two-lane turning lane, or that the road user stops before the turning maneuver of the ego-vehicle or leaves its lane, it is determined that there is no probability of a collision.
1. A system for improving assistance systems for lateral vehicle movements during at least two-lane turning maneuvers and turning maneuvers in which an adjacent lane ends before turning of an ego vehicle, the system comprising:
one electronic control unit, wherein the control unit is configured such that, successively or simultaneously:
a future turning maneuver of the ego vehicle is determined,
information about lane markings and a number of available lanes in a surrounding area in front of and next to the ego vehicle is detected, and
it is determined whether at least one further road user is in a relevant lane next to or behind the lane of the ego vehicle, and
upon determining that the at least one further road user is in the relevant lane next to or behind the lane of the ego vehicle, a future intended movement of the road user is determined from the information about the lane markings and the number of available lanes, and upon determining that the ego vehicle and the road user are in different lanes of an at least two-lane turning lane, that the road user stops before a turning maneuver of the ego vehicle, or that the road user leaves a lane of the road user, it is determined that there is no probability of a collision, wherein:
a learning algorithm is used in order to improve classifiers for determining the probability of the collision,
in training of the learning algorithm, chronologically continuous recordings of driving behavior of the ego vehicle, as well as surrounding area information, are input to a first classifier,
in training of the learning algorithm, the chronologically continuous recordings of driving behavior of the ego vehicle and chronologically continuous recordings of driving behavior of further vehicles, as well as the surrounding area information, are input to a second classifier,
an evaluation is made after completion of the turning maneuver of the ego vehicle,
the first classifier classifies the driving behavior of the ego vehicle, and
the second classifier classifies the driving behavior of the road user upon a determination by the first classifier that a lateral movement of the ego vehicle is present.
2. The system according to claim 1 , wherein upon determining that there is no probability of a collision, at least one of a warning to the driver or an intervention into a driving movement of the ego vehicle is suppressed.
3. The system according to claim 1 , wherein the surrounding area in front of the vehicle is detected as surrounding area information by a sensor system installed in the ego vehicle.
4. The system according to claim 3 , wherein the surrounding area information comprises information about at least one of a traffic situation, weather conditions, or a current road geometry.
5. The system according to claim 1 , wherein map information is used to make available information about the number and directions of lanes that are present.
6. A vehicle comprising the system according to claim 1 .
7. A method for improving assistance systems for lateral vehicle movements during at least two-lane turning maneuvers and turning maneuvers in which an adjacent lane ends before turning of an ego vehicle, the method comprising:
successively or simultaneously:
determining a future turning maneuver of the ego vehicle,
detecting information about lane markings and a number of available lanes in a surrounding area in front of and next to the ego vehicle, and
determining whether at least one further road user is in a relevant lane next to or behind the lane of the ego vehicle, and
upon determining that at least one further road user is in the relevant lane next to or behind the lane of the ego vehicle, a future driving task of the road user is determined from the information about the lane markings and the number of available lanes, and upon determining that the ego vehicle and the road user are in different lanes of an at least two-lane turning lane, or that the road user leaves a lane of the road user before the turning maneuver of the ego vehicle, it is determined that there is no probability of a collision, wherein:
a learning algorithm is used in order to improve classifiers for determining the probability of the collision,
in training of the learning algorithm, chronologically continuous recordings of driving behavior of the ego vehicle, as well as surrounding area information, are input to a first classifier,
in training of the learning algorithm, the chronologically continuous recordings of driving behavior of the ego vehicle and chronologically continuous recordings of driving behavior of further vehicles, as well as the surrounding area information, are input to a second classifier,
an evaluation is made after completion of the turning maneuver of the ego vehicle,
the first classifier classifies the driving behavior of the ego vehicle, and
the second classifier classifies the driving behavior of the road user upon a determination by the first classifier that a lateral movement of the ego vehicle is present.
8. A computer program product comprising a non-transitory computer readable medium having stored thereon program code which, when executed on a processor, carries out the acts of:
successively or simultaneously:
determining a future turning maneuver of the ego vehicle,
detecting information about lane markings and a number of available lanes in a surrounding area in front of and next to the ego vehicle, and
determining whether at least one further road user is in a relevant lane next to or behind the lane of the ego vehicle, and
upon determining that at least one further road user is in the relevant lane next to or behind the lane of the ego vehicle, a future driving task of the road user is determined from the information about the lane markings and the number of available lanes, and upon determining that the ego vehicle and the road user are in different lanes of an at least two-lane turning lane, or that the road user leaves a lane of the road user before the turning maneuver of the ego vehicle, it is determined that there is no probability of a collision, wherein:
a learning algorithm is used in order to improve classifiers for determining the probability of the collision,
in training of the learning algorithm, chronologically continuous recordings of driving behavior of the ego vehicle, as well as surrounding area information, are input to a first classifier,
in training of the learning algorithm, the chronologically continuous recordings of driving behavior of the ego vehicle and chronologically continuous recordings of driving behavior of further vehicles, as well as the surrounding area information, are input to a second classifier,
an evaluation is made after completion of the turning maneuver of the ego vehicle,
the first classifier classifies the driving behavior of the ego vehicle, and
the second classifier classifies the driving behavior of the road user upon a determination by the first classifier that a lateral movement of the ego vehicle is present.