Research on methods for constructing typical safety test scenarios in autonomous driving

Figures (17) Tables (8)

Figure 1.
Schematic of the test scenario construction process. ① selected dynamic and static scene parameters; ② filtered test scenarios.
Figure 2.
Multidimensional decomposition of scenario elements.
Figure 3.
Workflow of the variable-strength PICT combination testing method.
Figure 4.
Data collection area for the SPMD.
Figure 5.
PDF and CDF of SIC for extracted scenarios. (a) Car-following scenario. (b) Cut-in scenario.
Figure 6.
Random forest parameter training process. (a) The number of decision trees for the car-following scenario. (b) The number of feature variables for the car-following scenario. (c) The number of decision trees for the cut-in car scenario. (d) The number of feature variables for the cut-in car scenario.
Figure 7.
Ranking of feature importance in RF. (a) Car-following scenario. (b) Cut-in scenario.
Figure 8.
Spearman correlation analysis results. (a) Car-following scenario. (b) Cut-in scenario. The meanings of the numbers in the figure are as follows: ① ego vehicle speed, ② ego vehicle acceleration, ③ ego vehicle yaw rate, ④ leading vehicle speed, ⑤ leading vehicle acceleration, ⑥ relative longitudinal distance, ⑦ relative speed, ⑧ relative lateral distance, and ⑨ SIC.
Figure 9.
Probability distribution of basic dynamic parameters for Grand 3 car-following scenario samples. (a) Relative longitudinal distance. (b) Relative lateral distance. (c) Relative speed. (d) Leading vehicle speed. (e) Leading vehicle acceleration.
Figure 10.
Probability distribution of basic dynamic parameters for Grand 3 cut-in scenario samples. (a) Relative longitudinal distance. (b) Relative lateral distance. (c) Relative speed. (d) Leading vehicle speed. (e) Leading vehicle acceleration.
Figure 11.
Generation of car-following scenarios. (a) Original scenario. (b) PICT combined generated test scenario. (c) Filtered scenario.
Figure 12.
Generation of cut-in scenarios. (a) Original scenario. (b) PICT combined generated test scenario. (c) Filtered scenario.
Figure 13.
Elbow plot of clustering. (a) Car-following scenario. (b) Cut-in scenario.
Figure 14.
Silhouette coefficient plot for clustering. (a) Car-following scenario. (b) Cut-in scenario.
Figure 15.
Fifteen representative simulation test scenarios. (a) Car-following scenario. (b) Cut-in scenario.
Figure 16.
Simulation test results for car-following Scenario 1. (a) Ego vehicle speed. (b) Ego vehicle acceleration. (c) Relative distance.
Figure 17.
Simulation test results for cut-in Scenario 1. (a) Ego vehicle speed. (b) Ego vehicle acceleration. (c) Relative distance.

Static scenario elements			Parameter value
Natural environment elements	Weather conditions	Rainy	Small, medium, large
		Snowy	Small, medium, large
		Foggy	Small, medium, large
		Clear day	−
	Illumination condition		Day, night
Traffic environment elements	Number of lane lines		2, 3, 4
Traffic environment elements	Traffic flow		Low, medium, high

Table 1.

Parameter set for static scenario elements.

Screening index Screening condition

TTC TTC < 2 s

THW 20 m/s < V_x < 40 m/s THW < 0.01 × V_x+ 0.1

V_x ≥ 40 m/s THW ≤ 0.5 s

Table 2.
Scenario filtering criteria.
SIC range Safety impact level Number of scenarios

SIC ≤ 0.7539 Grand 3 2,214

0.7539 < SIC ≤ 0.8025 Grand 2 3,233

0.8025 < SIC ≤ 0.8395 Grand 1 7,923

SIC > 0.8395 Grand 0 5,209

Table 3.
Car-following scenario risk level classification.
SIC range Safety impact level Number of scenarios

SIC ≤ 0.7600 Grand 3 428

0.7600 < SIC ≤ 0.8004 Grand 2 724

0.8004 < SIC ≤ 0.8488 Grand 1 2,986

SIC > 0.8488 Grand 0 1,680

Table 4.
Cut-in scenario risk level classification.

Screening index	Screening condition
TTC	TTC < 2 s
THW	20 m/s < V_x < 40 m/s	THW < 0.01 × V_x+ 0.1
V_x ≥ 40 m/s	THW ≤ 0.5 s

SIC range	Safety impact level	Number of scenarios
SIC ≤ 0.7539	Grand 3	2,214
0.7539 < SIC ≤ 0.8025	Grand 2	3,233
0.8025 < SIC ≤ 0.8395	Grand 1	7,923
SIC > 0.8395	Grand 0	5,209

SIC range	Safety impact level	Number of scenarios
SIC ≤ 0.7600	Grand 3	428
0.7600 < SIC ≤ 0.8004	Grand 2	724
0.8004 < SIC ≤ 0.8488	Grand 1	2,986
SIC > 0.8488	Grand 0	1,680

Fitting parameter	Mean ($ \mu $)	Standard deviation ($ \sigma $)
Relative longitudinal distance	23.48	26.58
Relative lateral distance	0.03	1.03
Relative speed	−2.74	5.03
Leading vehicle speed	38.59	24.75
Leading vehicle acceleration	0.0011	0.0225

Table 5.

Normal distribution fit of key parameters for car-following scenarios.

Fitting parameter	Mean ($ \mu $)	Standard deviation ($ \sigma $)
Relative longitudinal distance	11.03	16.47
Relative lateral distance	1.52	0.02
Relative speed	−2.27	3.79
Leading vehicle speed	40.98	21.18
Leading vehicle acceleration	0.0012	0.0196

Table 6.

Normal distribution fit of key parameters for cut-in scenarios.

Scenario elements	Parameter value
Scenario elements	Car-following	Cut-in
Relative longitudinal distance	(1.5, 46.5) 1.5 m discrete	(3, 33) 3 m discrete
Relative lateral distance	0	(1, 3) 0.5 m/s discrete
Relative speed	(−10, 5) 1 m/s discrete	(−10, 5) 1 m/s discrete
Leading vehicle speed	(15, 60) 5 m/s discrete	(15, 60) 5 m/s discrete
Leading vehicle acceleration	0	0

Table 7.

Vehicle dynamic information elements set.

Scenario type	Scenario	WC	IC	NL	TF	LD (m)	LAD (m)	RS (m/s)	LS (m/s)	LA (m/s²)
Car-following	1	CD	N	6	H	19.5	0	0	30	0
	2	R-M	D	4	L	4.5	0	3	15	0
	3	CD	N	4	M	34.5	0	0	25	0
	4	S-S	D	6	M	10.5	0	−6	50	0
	5	CD	N	2	L	7.5	0	−9	35	0
	6	R-S	D	2	H	13.5	0	−10	55	0
	7	S-M	N	6	L	22.5	0	0	55	0
	8	R-M	N	4	H	4.5	0	−4	40	0
	9	F-S	D	6	H	1.5	0	−8	30	0
Cut-in	1	S-L	N	2	L	18	1.5	0	22	0
	2	CD	D	6	M	6	2.5	0	60	0
	3	R-M	N	4	M	15	1	−10	60	0
	4	S-S	N	6	H	24	2.5	−8	17	0
	5	R-L	D	2	H	12	3	−10	40	0
	6	F-L	D	4	L	33	2	0	20	0
WC, weather conditions; IC, illumination condition; NL, number of lane lines; TF, traffic flow; LD, relative longitudinal distance; LAD, relative lateral distance; RS, relative speed; LS, leading vehicle speed; LA, leading vehicle acceleration. In WC, CD represents clear day. If shown in the form 'a-b', a represents weather (R, S, F, for rainy, snowy, foggy, respectively), and b represents severity (S, M, L, for small, medium, large, respectively). In IC, D and N represent day and night, respectively. In TF, L, M, and H represent low, medium, and high, respectively.

Table 8.

Fifteen types of typical test scenarios.