Intersection right-turn trajectory planning method combined with digital traffic rules

Abstract

The application relates to a kind of intersection right-turn trajectory planning methods combined with digital traffic rules, comprising the following steps: the traffic rules involved in right-turn vehicle are divided into three categories: turning, yielding and speed limit; based on the expression of traffic rules formalized by metric time logic, the method combining machine learning and statistical analysis is used to quantify key parameters, and the traffic rules are digitized; combined with digital traffic rules, based on sampling and numerical optimization method, a right-turn trajectory planning model is constructed; a simulation platform is built to evaluate the right-turn trajectory planning model, select the optimal value of key parameters, and simulate and verify the effectiveness of the model based on the optimal value of key parameters. Compared with the prior art, the application has the advantages of high safety, strong rule compliance, etc.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving, and in particular to a method for planning right-turn trajectories at intersections that incorporates digital traffic rules. Background Technology

[0002] Traffic rules are crucial evidence for determining liability in accidents and a powerful tool for ensuring the safety of autonomous vehicles. During the development of autonomous driving, due to the lack of a mature technological system, the coexistence of human-driven and autonomous vehicles (human-machine hybrid driving) will continue for a long time. Autonomous vehicles must learn the traffic rules followed by human drivers to better cope with this complex traffic environment.

[0003] Existing traffic rules are designed for human drivers and are difficult for autonomous vehicles to adapt to. Current traffic rules often contain vague phrases, such as: "On roads without traffic signals, proceed only when it is safe and the flow of traffic is unobstructed," "When the green light is on, vehicles are permitted to proceed, but turning vehicles must not obstruct the passage of vehicles or pedestrians going straight," and "Right-turning vehicles traveling in the opposite direction must yield to left-turning vehicles." There are differences in perception between autonomous vehicles and human drivers. Human drivers can understand vague phrases like "ensuring safety," "not obstructing," and "yielding" through training and driving experience, but autonomous vehicles require clearer and more comprehensive explanations.

[0004] The ambiguity of traffic rules leads to a lack of consideration for traffic rules in trajectory planning methods for autonomous vehicles. Traditional trajectory planning methods aim to maximize comfort and efficiency while avoiding collisions. Even when considering the impact of traffic rules, the key parameters of the rules are obtained through human definition or statistical analysis, which cannot meet the needs of autonomous vehicles to operate safely and compliantly in human-machine co-driving environments. Summary of the Invention

[0005] The purpose of this invention is to provide a method for planning right-turn trajectories at intersections that combines efficiency, comfort, and safety with digital traffic rules.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for planning right-turn trajectories at intersections that incorporates digital traffic rules includes the following steps:

[0008] Step 1) Divide the traffic rules for right-turning vehicles into three categories: turning, yielding, and speed limits;

[0009] Step 2) Based on the formal expression of traffic rules using Metric Time Logic (MTL), key parameters are quantified using a combination of machine learning and statistical analysis to digitize traffic rules;

[0010] Step 3) Combining digital traffic rules, construct a right-turn trajectory planning model based on sampling and numerical optimization methods;

[0011] Step 4) Build a simulation platform, evaluate the right-turn trajectory planning model, select the optimal values ​​of key parameters, and verify the effectiveness of the model based on the optimal values ​​of key parameters through simulation.

[0012] Step 2) includes the following steps:

[0013] Step 21) Based on the time-based metric logic, convert the natural language of each traffic rule into a propositional expression, determine the key parameters, and decompose each expression into the smallest atomic proposition that can be combined with the key parameters.

[0014] Step 22) Use a combination of machine learning and statistical analysis to quantify key parameters.

[0015] The propositional expressions for each of the traffic rules are as follows:

[0016] i) Turning:

[0017] ii) Give way:

[0018] iii) Speed ​​limit:

[0019] Among them, T rightturn Indicates whether to perform a normal right turn; T rightway Indicates whether to yield; T speedlimitation Indicates whether the speed limit is observed; T indicates yes; F indicates no; locationRangeTurn indicates whether the vehicle is within the turning range; SafeViolation indicates whether the vehicle is within the safety constraint; LocationRange indicates whether the vehicle is within the deceleration range; SpeedViolation indicates whether the vehicle is within the speed limit constraint.

[0020] The specific steps of decomposing each expression into the smallest atomic proposition that can be combined with key parameters are as follows:

[0021] Turning range (LocationRangeTurn) uses the starting position of the turn (x). turnstop * End of turn position x turnend *Indicates: Turning Safety Threshold (SafeViolation) and Time-of-Use Safety Threshold (TTC) * Relative distance x relative * Indicates; the deceleration range LocationRange uses the deceleration start position x. stop * End of turn position x turnend * SpeedViolation uses v to limit speeds when turning. limit * This indicates that the smallest atomic proposition is represented as:

[0022]

[0023]

[0024]

[0025]

[0026]

[0027] in, This indicates that the turn is happening every moment from the start of the turn t1 to the end of the turn t2; distance(Ego,Stopline) represents the distance from the vehicle to the stop line of the approach lane; distance(Ego,Stoplineexit) represents the distance from the vehicle to the stop line of the exit lane; Tgt represents other vehicles; ttc(Ego,Tgt) represents the TTC value between the vehicle and other vehicles; distance(Ego,Tgt) represents the relative distance between the vehicle and other vehicles; distancey(Ego,Tgt) represents the longitudinal relative distance between the vehicle and other vehicles; distancex(Ego,Tgt) represents the lateral relative distance between the vehicle and other vehicles. This indicates that the deceleration begins at t3 and ends at t2, occurring every moment. This indicates that the event occurred from the start of the turn (t1) to the end of the turn (t2); v x (Ego) represents the speed of the vehicle at a certain moment.

[0028] The key parameters are determined from both time and space perspectives. The time index is represented by Extended Time to Collision (ETTC), and the space index is represented by relative distance.

[0029] Step 22) specifically involves:

[0030] Step 221) For the time index ETTC, machine learning methods are used for quantification: the severity of conflict based on ETTC is used to represent the security level, divided into three categories: severe, minor, and potential conflict. After manual screening and removal of outliers by local outliers, k-means++ is used to cluster the maximum Jerk, relative velocity, and relative distance into three categories. The ETTC range of each category is selected to constitute the security boundary of that category, which is the TTC. * ;

[0031] Step 222) Quantify spatial indicators using statistical analysis methods: Considering the characteristics of traffic operation, select typical values, such as the 15th and 85th percentiles, and the largest proportion, as x. relative * ; Analyze the distance between right-turning vehicles and the stop line of the approach lane during serious conflicts, and select the location with the highest percentage as x. turnstop * ; Analyze the changes in speed with vehicle position at all times of conflict, and use the significant deceleration point as x. stop * ;v limit * Select 30km / h; the turn is considered complete after the turning vehicle leaves the stop line of the exit lane, and this is used to determine x. turnend * .

[0032] Step 3) includes the following steps:

[0033] Step 31) Determine different trajectory sampling methods for each type of traffic rule, where speed limit rules are incorporated into right-of-way and turning rules;

[0034] Step 32) Based on the sampling terminal position, use a fifth-order polynomial to fit and optimize the trajectory from the sampling start point to the end point, and construct a right-turn trajectory planning model;

[0035] Step 33) Optimal trajectory selection: Select the optimal trajectory based on a comprehensive consideration of three indicators: safety, efficiency, and comfort.

[0036] In step 31), all sampling is completed in the Frenet coordinate system. In the Frenet coordinate system, the terminal state of the sampling is represented by [s ds dds l dl ddl], where s is the longitudinal displacement, ds is the longitudinal velocity, dds is the longitudinal acceleration, l is the lateral displacement, dl is the lateral velocity, and ddl is the lateral acceleration.

[0037] In step 31), three terminal states x are specified. Ego The sampling method of (k), where,

[0038] The sampling method for turning is: x Ego(k)=[NaN ds turn [0 l 0 0], the sampling time T is calculated by uniformly sampling 6 points between 2s and 4s;

[0039] The sampling method for yielding is: x Ego (k)=[s stop [0 0 0 0 0], where the sampling time T is the time taken to obtain the trajectory with optimal smoothness using the Minimum-Jerk method;

[0040] The sampling method for driving outside the turning range is: x Ego (k)=[NaN ds normal [0 l 0 0], the sampling time T is calculated by uniformly sampling 6 points between 2s and 4s;

[0041] Where NaN indicates that there are no requirements for the terminal state of s; ds turn This represents the sampling of the terminal speed during the turn, in the range of 0 to v. limit * Ten points are sampled evenly between lanes; l represents the lateral sampling position, selected between 0 and lanewidth, where lanewidth represents the width of a lane; s stop To determine the longitudinal position of the sampling terminal, vehicles must stop before the stop line when sampling is being conducted; ds normal This represents the sampling of the terminal speed when driving outside the turning range, in the range of 0 to V. limitnormal Ten points were sampled evenly between them, V limitnormal This indicates the speed limit outside the turning range.

[0042] Step 33) specifically refers to:

[0043] Safety metrics use the collision probability p c As evaluation indicators, comfort-related indicators use the cumulative Jerk values ​​in both the horizontal and vertical directions, while efficiency-related indicators use the speed difference between the average value of the sampled trajectory and the speed limit. After normalizing the three types of indicators, they are summed, and the trajectory with the smallest value is selected as the optimal sampled trajectory. Where p... c Represented using the Gaussian function:

[0044]

[0045] Where ΔR = RR min R min =min[L / 2,W / 2], where R is the distance between the other car and the car itself, L is the length of the car itself, W is the width of the car, σ is the variance of the Gaussian function, and R max This represents the maximum distance at which a collision is possible.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] (1) This invention improves the safety of right turns for autonomous vehicles by digitizing traffic rules and combining them with trajectory planning methods;

[0048] (2) This invention proposes a formalized and quantitative method for digitizing traffic rules, which makes up for the shortcomings of existing traffic rules that are vague and unclear, and is more suitable for autonomous driving scenarios;

[0049] (3) This invention embeds digital traffic rules into the trajectory planning sampling method, which makes up for the lack of consideration of traffic rules in the existing trajectory planning model, making autonomous vehicles more adaptable to human-machine mixed driving environment. Attached Figure Description

[0050] Figure 1 This is a flowchart of the method of the present invention;

[0051] Figure 2 This is a schematic diagram of a trajectory planning method combined with digital traffic rules in one embodiment;

[0052] Figure 3 This is a schematic diagram showing the changes in collision time ETTC between the trajectory planning method considering rules and the trajectory planning method not considering rules in one embodiment of the present invention.

[0053] Figure 4 This is a schematic diagram illustrating the collision probability changes of the trajectory planning method considering rules and the trajectory planning method not considering rules in one embodiment of the present invention.

[0054] Figure 5 This is a schematic diagram illustrating the Jerk variation of the trajectory planning method considering rules and the trajectory planning method not considering rules in one embodiment of the present invention;

[0055] Figure 6 This is a schematic diagram illustrating the speed difference between the trajectory planning method that considers rules and the trajectory planning method that does not consider rules, according to one embodiment of the present invention. Detailed Implementation

[0056] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0057] This embodiment provides a method for planning right-turn trajectories at intersections that incorporates digital traffic rules, such as... Figure 1 As shown, it includes the following steps:

[0058] Step 1) Divide the traffic rules for right-turning vehicles into three categories: turning, yielding, and speed limits;

[0059] Step 2) Based on the formal expression of traffic rules using Metric Time Logic (MTL), key parameters are quantified using a combination of machine learning and statistical analysis to digitize traffic rules;

[0060] Specifically, step 2) includes the following steps:

[0061] Step 21) Based on the time-based metric logic, the natural language of each traffic rule is converted into a propositional expression, key parameters are determined, and each expression is decomposed into the smallest atomic proposition that can be combined with the key parameters.

[0062] The propositions for each traffic rule are expressed in the following forms:

[0063] i) Turning:

[0064] ii) Give way:

[0065] iii) Speed ​​limit:

[0066] Among them, T rightturn Indicates whether to perform a normal right turn; T rightway Indicates whether to yield; T speedlimitation This indicates whether the speed limit is being followed; T indicates yes; F indicates no; LocationRangeTurn indicates whether the vehicle is within the turning range; SafeViolation indicates whether a safety constraint has been violated; LocationRange indicates whether the vehicle is within the deceleration range; SpeedViolation indicates whether a speed limit constraint has been violated.

[0067] In this embodiment, key parameters are determined from both temporal and spatial perspectives. The temporal metric uses Extended Time to Collision (ETTC), while the spatial metric uses relative distance. The turning range (LocationRangeTurn) uses the turning start position x... turnstop * End of turn position x turnend * Indicates: Turning Safety Threshold (SafeViolation) and Time-of-Use Safety Threshold (TTC) * relative distance x relative * Indicates; the deceleration range LocationRange uses the deceleration start position x. stop *End of turn position x turnend * SpeedViolation uses v to limit speeds when turning. limit * express.

[0068] Further decompose each expression into the smallest atomic proposition that can be combined with key parameters:

[0069]

[0070]

[0071]

[0072]

[0073] in, This indicates that the turn is happening every moment from the start of the turn t1 to the end of the turn t2; distance(Ego,Stopline) represents the distance from the vehicle to the stop line of the approach lane; distance(Ego,Stoplineexit) represents the distance from the vehicle to the stop line of the exit lane; Tgt represents other vehicles; ttc(Ego,Tgt) represents the TTC value between the vehicle and other vehicles; distance(Ego,Tgt) represents the relative distance between the vehicle and other vehicles; distancey(Ego,Tgt) represents the longitudinal relative distance between the vehicle and other vehicles; distancex(Ego,Tgt) represents the lateral relative distance between the vehicle and other vehicles. This indicates that the deceleration begins at t3 and ends at t2, occurring every moment. This indicates that the event occurred from the start of the turn (t1) to the end of the turn (t2); v x (Ego) represents the speed of the vehicle at a certain moment.

[0074] Step 22) Use a combination of machine learning and statistical analysis to quantify key parameters.

[0075] Step 221) For the time index ETTC, machine learning methods are used for quantification: the severity of conflict based on ETTC is used to represent the security level, divided into three categories: severe, minor, and potential conflict. After manual screening and removal of outliers by local outliers, k-means++ is used to cluster the maximum Jerk, relative velocity, and relative distance into three categories. The ETTC range of each category is selected to constitute the security boundary of that category, which is the TTC. * .

[0076] Step 222) Quantify spatial indicators using statistical analysis methods: Considering the characteristics of traffic operation, select typical values, such as the 15th and 85th percentiles, and the largest proportion, as x. relative * ; Analyze the distance between right-turning vehicles and the stop line of the approach lane during serious conflicts, and select the location with the highest percentage as x. turnstop * ; Analyze the changes in speed with vehicle position at all times of conflict, and use the significant deceleration point as x. stop * ;v limit * According to a certain regulation, a speed of 30 km / h is selected; the turn is considered complete after the turning vehicle leaves the stop line of the exit lane, and this is used to determine x. turnend * .

[0077] This embodiment is based on a day's trajectory data of an intersection in a certain city, from which 1584 right-turn trajectories were extracted and analyzed.

[0078] For the time metric, the ETTC range for each category is selected to constitute the safety boundary for each category. 0.98s, 139s, and 1.91s are obtained as the thresholds for the three types of conflicts. Further rounding, 1s, 1.4s, and 2s are selected as the TTC values. * Alternative values.

[0079] For spatial indicators, the relative distances at the occurrence of severe, minor, and potential conflicts were statistically analyzed. The 15th and 85th percentiles were 4.07m, 18.06m, 5.01m, 22.35m, 5.84m, and 17.38m, respectively. When braking measures were taken after a severe conflict and the ETTC initiation time was greater than 2 seconds, the relative distance for most events (85%) was less than 34.1m. Therefore, 4m, 5m, 6m, 17m, 18m, 22m, and 34m were used as x relative * The alternative is to statistically analyze the speed changes with position at all collisions occurring before the approach lane, and use the distance from the significant deceleration point to the stop line of the approach lane as x. stop * x stop * =27m. Serious conflicts occur most frequently within 10m of the stop line at the approach lane, i.e., x turnstop * =10m. The turn is considered complete when the vehicle has moved away from the exit lane stop line, i.e., when the distance from the vehicle to the exit lane stop line is 0m. turnend * =0m. v limit * Set to 30km / h.

[0080] Step 3) as Figure 2 As shown, a right-turn trajectory planning model is constructed by combining digital traffic rules and based on sampling and numerical optimization methods.

[0081] Specifically, step 3) includes the following steps:

[0082] Step 31) Determine different trajectory sampling methods for each type of traffic rule, where speed limit rules are incorporated into yielding and turning rules.

[0083] In this embodiment, all sampling is completed in the Frenet coordinate system. In the Frenet coordinate system, the terminal state of the sampling is represented by [s ds dds l dl ddl], where s is the longitudinal displacement, ds is the longitudinal velocity, dds is the longitudinal acceleration, l is the lateral displacement, dl is the lateral velocity, and ddl is the lateral acceleration.

[0084] Then, three terminal states x are specified. ego The sampling method for (k) is as follows:

[0085] The sampling method for turning is: x Ego (k)=[NaN ds turn [0 l 0 0], the sampling time T is calculated by uniformly sampling 6 points between 2s and 4s;

[0086] The sampling method for yielding is: x Ego (k)=[s stop [0 0 0 0 0], where the sampling time T is the time taken to obtain the trajectory with optimal smoothness using the Minimum-Jerk method;

[0087] The sampling method for driving outside the turning range is: x Ego (k)=[NaN ds normal [0 l 0 0], the sampling time T is calculated by uniformly sampling 6 points between 2s and 4s;

[0088] NaN indicates that there are no requirements regarding the terminal state of s; ds turn The sampling point represents the terminal speed during the turn, with 10 points evenly sampled between 0 and 30 km / h; 'l' represents the lateral position sampling, selected between 0 and 'lanewidth', where 'lanewidth' represents the width of a lane, set to 3.2m; 's' represents the lateral position sampling. stop To determine the longitudinal position of the sampling terminal, vehicles need to stop before the stop line during sampling; therefore, s stop The position of the stop line at the entrance; ds normal To sample the terminal speed when driving outside the turning range, 10 points were uniformly sampled between 0 and 60 km / h.

[0089] Step 32) Based on the sampling terminal position, use a fifth-order polynomial to fit and optimize the trajectory from the sampling start point to the end point, and construct a right-turn trajectory planning model;

[0090] Step 33) Optimal trajectory selection: Select the optimal trajectory based on a comprehensive consideration of three indicators: safety, efficiency, and comfort.

[0091] Specifically, safety metrics use the collision probability p c As evaluation indicators, comfort-related indicators use the cumulative Jerk values ​​(horizontal and longitudinal) as evaluation metrics, while efficiency-related indicators use the speed difference between the average value of the sampled trajectory and the speed limit. After normalizing the three evaluation metrics, they are summed, and the trajectory with the smallest value is selected as the optimal sampled trajectory, where p... c Represented using the Gaussian function:

[0092]

[0093] Where ΔR = RR min R min =min[L / 2,W / 2], where R is the distance between the other car and the car itself, L is the length of the car itself, W is the width of the car, σ is the variance of the Gaussian function, and R max This represents the maximum distance at which a collision occurs. To avoid redundant calculations, the maximum range is set to R. max If R≥R max If R ≤ R min If this happens, a collision will definitely occur. In this embodiment, the maximum range R is... max The width of the two lanes is set at 6.4m.

[0094] Step 4) Build a simulation platform, evaluate the right-turn trajectory planning model, select the optimal values ​​of key parameters, and verify the effectiveness of the model based on the optimal values ​​of key parameters through simulation.

[0095] Specifically, MATLAB's Automated Driving Toolbox, Sensor Fusion, Tracking Toolbox, and Navigation Toolbox were used for scene generation, map creation, sensor data fusion, and simulation. An interactive vehicle was set up based on real trajectory data to simulate a right-turn interaction scenario. The simulation step size was set to 0.1 seconds. The trajectory planning method considering rules in this invention was evaluated using the metrics proposed in step 33) to determine the TTC. * x relatove * The optimal parameters were selected, and the effectiveness of the constructed algorithm was verified by simulation based on the optimal parameters.

[0096] For TTC * x relative * The alternative parameters were evaluated using the indicators proposed in step 33), and TTC was found to be... * =2s,x relative * =4m is the optimal parameter.

[0097] The proposed method is based on optimal parameter evaluation. Regarding vehicle safety, the rule-based method achieves a higher ETTC value and a lower collision probability compared to the rule-independent method, indicating better safety. For example... Figure 3 , Figure 4 As shown, the probability of collision during turning decreased from 0.097 to 0, and the ETTC increased from 0.8s to 1.57s, indicating that driving is safer.

[0098] In terms of operational comfort, such as Figure 5 As shown, due to yielding and deceleration, the Jerk value fluctuates slightly compared to the method that does not consider rules before turning, but still remains within the comfortable range of [-3.5, 1.2] m / s. 3 .

[0099] In terms of operational efficiency, such as Figure 6 As shown, the method of the present invention that takes rules into account has a larger speed difference than the method that does not take rules into account, sacrificing some efficiency to ensure safety.

[0100] Based on the above, the intersection right-turn trajectory planning method of the present invention, which incorporates digital traffic rules, has proven to be safer and more compliant with the rules.

[0101] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for planning right-turn trajectories at intersections based on digital traffic rules, characterized in that, Includes the following steps: Step 1) Divide the traffic rules for right-turning vehicles into three categories: turning, yielding, and speed limits; Step 2) Based on the formalized expression of traffic rules according to the time logic of measurement, use a combination of machine learning and statistical analysis to quantify key parameters and digitize traffic rules; Step 3) Combining digital traffic rules, construct a right-turn trajectory planning model based on sampling and numerical optimization methods; Step 4) Build a simulation platform, evaluate the right-turn trajectory planning model, select the optimal values ​​of key parameters, and verify the effectiveness of the model through simulation based on the optimal values ​​of key parameters; Step 2) includes the following steps: Step 21) Based on the time-based metric logic, convert the natural language of each traffic rule into a propositional expression, determine the key parameters, and decompose each expression into the smallest atomic proposition that can be combined with the key parameters; Step 22) Quantify key parameters using a combination of machine learning and statistical analysis; The propositional expressions for each of the traffic rules are as follows: i) Turning: ii) Yield: iii) Speed ​​limit: in, Indicates whether a normal right turn operation should be performed; Indicate whether to yield; Indicate whether you are complying with the speed limit; It means yes; Indicates no; Indicates whether it is within the turning range; Indicate whether safety constraints have been violated; Indicates whether it is within the deceleration range; Indicates whether a speed limit has been violated; The specific steps of decomposing each expression into the smallest atomic proposition that can be combined with key parameters are as follows: Turning range Use the starting position of the turn End of turn Indicates the turning safety threshold. Use time safety threshold Relative distance Indicates the deceleration range. Use deceleration starting position End of turn Speed ​​limits during turns use The smallest atomic proposition is represented as: : ) : ) : ) > in, Indicates starting from the turn. Until the end of the turn It happens every moment; Indicates the distance from the vehicle to the stop line of the approach lane; Indicates the distance from the vehicle to the stop line at the exit lane; Indicates his car; This indicates the TTC value between the vehicle and other vehicles; Indicates the relative distance between your vehicle and other vehicles; Indicates the longitudinal relative distance between your vehicle and other vehicles; Indicates the lateral relative distance between your vehicle and other vehicles; Indicates starting from deceleration Until the end of the turn It happens every moment; Indicates starting from the turn. Until the end of the turn It happened; It indicates the speed of the vehicle at a certain moment.

2. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 1, characterized in that, The key parameters are determined from both time and space perspectives. The time index is represented by extended collision time, and the space index is represented by relative distance.

3. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 2, characterized in that, Step 22) specifically involves: Step 221) For the time index ETTC, machine learning methods are used for quantification: the severity of the collision based on ETTC is used to represent the safety level, divided into three categories: severe, minor, and potential collision. After manual screening and removal of outliers by local outliers, k-means++ is used to cluster the maximum Jerk, relative velocity, and relative distance into three categories. The extended collision time range of each category is selected to constitute the safety boundary of that category, i.e., the... ; Step 222) Quantify spatial indicators using statistical analysis methods: Considering the characteristics of traffic operation, select typical values ​​as... ; Analyze the distance between right-turning vehicles and the stop line of the approach lane during serious conflicts, and select the location with the highest percentage as... ; Analyze the changes in speed with vehicle position at all times of conflict to identify significant deceleration points as... ; Select 30km / h; the turn is considered complete after the turning vehicle leaves the stop line of the exit lane. .

4. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 1, characterized in that, Step 3) includes the following steps: Step 31) Determine different trajectory sampling methods for each type of traffic rule, where speed limit rules are incorporated into right-of-way and turning rules; Step 32) Based on the sampling terminal position, use a fifth-order polynomial to fit and optimize the trajectory from the sampling start point to the end point, and construct a right-turn trajectory planning model; Step 33) Optimal trajectory selection: Select the optimal trajectory based on a comprehensive consideration of three categories of indicators: safety, efficiency, and comfort.

5. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 4, characterized in that, In step 31), all sampling is performed in the Frenet coordinate system. In the Frenet coordinate system, the terminal state of the sample is represented by […]. ] indicates that, For longitudinal displacement, For longitudinal velocity, For longitudinal acceleration, This is a lateral displacement. For lateral velocity, This is lateral acceleration.

6. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 4, characterized in that, In step 31), three terminal states are specified. The sampling method, among which, The sampling method for turning is: = The sampling time T is calculated by uniformly sampling 6 points between 2s and 4s. The sampling method for yielding is as follows: = The sampling time T is the time required to obtain the trajectory with optimal smoothness using the Minimum-Jerk method; The sampling method for driving outside the turning range is as follows: = The sampling time T is calculated by uniformly sampling 6 points between 2s and 4s. in, Indicates to There are no requirements regarding the terminal status; This represents the sampling of the terminal speed during the turn, ranging from 0 to... Ten points were sampled evenly between them; The sampling indicates the lateral position, in 0 and Choose between them Indicates the width of a lane; The longitudinal position of the sampling terminal is such that vehicles must stop before the stop line when sampling is performed; This represents the sampling of the terminal speed when driving outside the turning range, in the range of 0 to... Ten points were sampled evenly between them. This indicates the speed limit outside the turning range.

7. The intersection right-turn trajectory planning method combining digital traffic rules according to claim 4, characterized in that, Step 33) specifically involves: Safety metrics use collision probability As evaluation indicators, comfort-related indicators use the cumulative Jerk values ​​in both horizontal and vertical directions, while efficiency-related indicators use the speed difference between the average value of the sampled trajectory and the speed limit. After normalizing the three types of indicators, they are summed, and the trajectory with the smallest value is selected as the optimal sampled trajectory. Represented using the Gaussian function: in = , = , The distance between his car and his own. L The length of the vehicle. W For vehicle width, Let V be the variance of the Gaussian function. This represents the maximum distance at which a collision is possible.

Images