Computer-implemented method for determining similarity values of traffic scenarios
A method for determining similarity values of traffic scenarios based on movement profiles of traffic participants in the traffic scenarios includes: generating, by a test device, at least one movement profile of a traffic participant using measurement data, wherein the at least one movement profile comprises at least one movement element, and wherein a new segment in the at least one movement profile begins with a change of a movement element; generating, by the test device, a sequence of the at least one movement profile for an ego vehicle and/or a fellow vehicle of a traffic scenario, wherein a sequence change is generated based on a segment being changed for a movement element; determining, by the test device, a measure of similarity based on movement profiles from at least two traffic scenarios by comparing respective sequences created from the movement profiles; and providing the measure of similarity.
1. A method, comprising:
obtaining, by a test device, a plurality of sequences corresponding to a plurality of traffic scenarios, wherein each sequence of the plurality of sequences contains a plurality of segments, wherein each segment of the plurality of segments of a respective sequence contains a combination of parameters corresponding to a movement profile of a vehicle, and wherein consecutive segments of a respective sequence of the plurality of sequences are different from one another with respect to at least one parameter value;
determining, by the test device, measures of similarity between traffic scenarios of the plurality of traffic scenarios by comparing respective sequences of the plurality of sequences; and
selecting, by the test device, a subset of traffic scenarios of the plurality of traffic scenarios to be used for virtual testing of one or more components of an autonomous vehicle based on the determined measures of similarity;
wherein the combination of parameters includes two or more of the following:
a lateral movement parameter indicative of a lane change being performed;
a longitudinal movement parameter indicative of vehicle speed;
a distance parameter indicative of distances of one or more other vehicles relative to an ego vehicle; or
a position parameter indicative of positions of one or more other vehicles relative to an ego vehicle.
2. The method according to claim 1 , wherein each sequence includes a chronological progression of segments, including changes in at least one parameter of the combination of parameters from one segment to a next segment.
3. The method according to claim 1 , wherein a new segment is generated based on a change being determined in at least one parameter of the combination of parameters.
4. The method according to claim 1 , wherein a respective measure of similarity is determined by comparing respective sequences using a normalized weighted edit distance, wherein the respective measure of similarity between a first sequence and a second sequence is calculated as a minimum sum of edit operations for converting the first sequence into the second sequence, wherein the edit operations include delete and/or move and/or insert operations.
5. The method according to claim 4 , wherein the calculation of a minimum sum of edit operations takes into account a cost associated with carrying out a delete operation or an insert operation.
6. The method according to claim 5 , wherein the following difference equation is defined for the edit distance:
D
(
i
,
j
)
=
min
{
D
(
i
-
1
,
j
)
+
d
(
S
1
,
i
-
1
,
S
1
,
i
)
+
λ
Deletion
D
(
i
,
j
-
1
)
+
d
(
S
2
,
j
-
1
,
S
2
,
j
)
+
λ
Insertion
D
(
i
-
1
,
j
-
1
)
+
d
(
S
1
,
i
-
1
,
S
2
,
j
-
1
)
Replacement
where S n,i and S n,j correspond to respective states of respective sequences, (i,) is an element of a table in which intermediate results of calculating costs of edit operations between two states i and j of the two sequences are stored, and λ corresponds to costs that arise from carrying out a delete or insert operation.
7. The method according to claim 6 , wherein the following function is used for calculating the costs of edit operations between the two states Si and S′j of the two sequences:
d ( Si,S′j )= wMa*dMa ( Si,S′j )+ wGe*dGe ( Si,S′j )+ wDi*dDi ( Si,S′j )+ wPo*dPo ( Si,S′j )
where the function is a weighted sum of the costs that are generated between individual classes of two respective states, and where dMa, dGe, dDi, and dPo correspond to respective parameters of the combination of parameters.
8. A test device, comprising:
a processor; and
a memory having processor-executable instructions stored thereon;
wherein the processor is configured to execute the processor-executable instructions to cause the test device to perform the following:
obtaining a plurality of sequences corresponding to a plurality of traffic scenarios, wherein each sequence of the plurality of sequences contains a plurality of segments, wherein each segment of the plurality of segments of a respective sequence contains a combination of parameters corresponding to a movement profile of a vehicle, and wherein consecutive segments of a respective sequence of the plurality of sequences are different from one another with respect to at least one parameter value;
determining measures of similarity between traffic scenarios of the plurality of traffic scenarios by comparing respective sequences of the plurality of sequences; and
selecting, by the test device, a subset of traffic scenarios of the plurality of traffic scenarios to be used for virtual testing of one or more components of an autonomous vehicle based on the determined measures of similarity;
wherein the combination of parameters includes two or more of the following:
a lateral movement parameter indicative of a lane change being performed;
a longitudinal movement parameter indicative of vehicle speed;
a distance parameter indicative of distances of one or more other vehicles relative to an ego vehicle; or
a position parameter indicative of positions of one or more other vehicles relative to an ego vehicle.
9. The test device according to claim 8 , wherein the test device is formed by a control unit and is configured to use scenarios based on virtual tests, real tests, and/or measurement data.
10. A non-transitory computer-readable medium having processor-executable instructions stored thereon, wherein the processor-executable instructions, when executed, facilitate:
obtaining, by a test device, a plurality of sequences corresponding to a plurality of traffic scenarios, wherein each sequence of the plurality of sequences contains a plurality of segments, wherein each segment of the plurality of segments of a respective sequence contains a combination of parameters corresponding to a movement profile of a vehicle, and wherein consecutive segments of a respective sequence of the plurality of sequences are different from one another with respect to at least one parameter value;
determining, by the test device, measures of similarity between traffic scenarios of the plurality of traffic scenarios by comparing respective sequences of the plurality of sequences; and
selecting, by the test device, a subset of traffic scenarios of the plurality of traffic scenarios to be used for virtual testing of one or more components of an autonomous vehicle based on the determined measures of similarity;
wherein the combination of parameters includes two or more of the following:
a lateral movement parameter indicative of a lane change being performed;
a longitudinal movement parameter indicative of vehicle speed;
a distance parameter indicative of distances of one or more other vehicles relative to an ego vehicle; or
a position parameter indicative of positions of one or more other vehicles relative to an ego vehicle.