A safe travel route planning method for elderly drivers considering travel costs

Abstract

The present invention relates to a method for planning safe travel routes for elderly drivers that takes travel costs into account, belonging to the field of road traffic safety technology. The method primarily comprises: Step 1: Collecting relevant data about elderly drivers; Step 2: Constructing a cost function for elderly drivers' safety and reliability; Step 3: Constructing a cost function for elderly drivers' driving confidence; Step 4: Constructing a cost function for elderly drivers' driving ability; Step 5: Constructing a cost function for road impedance; and Step 6: Planning safe travel routes for elderly drivers. The present invention can provide route solutions for elderly drivers' driving trips, helping them achieve safe travel.

Description

Technical Field

[0001] The present invention belongs to the field of traffic safety, and more specifically relates to a method for planning safe travel routes for elderly drivers taking travel costs into consideration. Background Art

[0002] To investigate driver safety and reliability in different road traffic environments, a safety-focused route planning scheme for elderly drivers was constructed. The primary purpose of the simulation experiment was to collect data on various driver behavioral characteristics under different driving environments and tasks, including psychological, eye movement, and operational behavior. The data collected and analyzed included psychological characteristics such as galvanic skin changes and heart rate increases; eye movement characteristics such as gaze duration, saccade amplitude, and blink frequency; and operational behavior characteristics such as speed and accelerator pedal depth. The data analyzed the differences between drivers of different ages in different driving environments and explored the relationship between driver behavioral characteristics and intersection conflicts. Summary of the Invention

[0003] The present invention can provide a route plan for elderly people's driving and travel, helping them to travel safely.

[0004] In order to achieve the above object, the present invention is implemented by adopting the following technical solutions: the method comprises:

[0005] Step 1: Collect relevant data of elderly drivers;

[0006] Step 2: Construct a cost function for the safety and reliability of elderly drivers;

[0007] Step 3: Construct a cost function for elderly drivers’ driving confidence;

[0008] Step 4: Construct a cost function for the driving ability of elderly drivers;

[0009] Step 5: Construct the cost function of road impedance;

[0010] Step 6: Plan safe travel routes for elderly drivers.

[0011] Furthermore, in step 1, data collection related to elderly drivers is carried out by recruiting elderly drivers with a certain driving age and experience and without diseases that affect driving ability, obtaining the drivers' driving behavior characteristic parameters, psychological and physiological characteristics, and operational behavior data through driving simulation experiments, and conducting a questionnaire survey on the elderly drivers' driving confidence in different driving scenarios.

[0012] Furthermore, in step 2, the cost function of the safety reliability of elderly drivers is constructed; the reliability level of the driver in different scenarios can be expressed as the weighted sum of the average conflict complexity under the three conflict forms, and the cost function f of the driver's safety reliability is considered. r (n) Such as:

[0013]

[0014] Where k is the weight coefficient, which is used to calculate the driving task according to the driving scenario; f r The larger (n) is, the greater the complexity of the intersection conflict is and the lower the driver reliability level is.

[0015] Furthermore, in step 3, the cost function of elderly drivers' driving confidence is constructed; the lack of driving confidence on the road is obtained as:

[0016] f c =k1N1+k2N2+k3N3+k4N4

[0017] N i is the statistical number of different scenarios, k i is the weight coefficient of different scenarios; k is the weight coefficient of confidence loss in different scenarios; N1, N2, and N3 are the number of three low-confidence scenarios respectively; N4 is the number of scenarios other than the three low-confidence traffic scenarios, including driving scenarios such as expressways, and is counted according to the characteristics of intersections and road sections.

[0018] Furthermore, the cost function of the driving ability in step 4 is as follows:

[0019] A comprehensive evaluation of elderly drivers' driving ability is conducted based on their self-assessment scores on the health status, driving ability, driving skills, and confidence level scales. Different weights are assigned to each module in the total score calculation according to the number of questions and corresponding scores in each submodule of the designed scale. The calculation method is shown in the following formula:

[0020]

[0021] Where wi is the weight of the ith submodule, ti is the score of the ith submodule, i = 1, 2, 3, 4, and ttotal is the total score of the test scale.

[0022] The formula for calculating the total score of the subjective assessment of driving ability is as follows:

[0023]

[0024] Where sj is the total score of the subjective evaluation of driving ability in the jth test, and the total score range is [0,100]; sij is the score of the i-th submodule in the j-th test, i = 1, 2, 3, 4, j∈N*; ttotal is the total score of the test scale;

[0025] The total score of the subjective evaluation of the elderly driver's driving ability is used to obtain the comprehensive evaluation result of their driving ability. The total score range is between 0 and 100 points, and each 10-point interval within the range is a level assessment, corresponding to (0-60) poor, (60-70) medium, (70-80) good, (80-90) excellent, and (90-100) excellent. The test results divide the subject's driving ability into different levels;

[0026] The driving ability cost function of elderly drivers is expressed as the weighted sum of the scores of the four modules: health status, driving ability, driving skills, and driving ability confidence level, that is:

[0027] f a =w1a1+w2a2+w3a3+w4a4

[0028] where a1 is the score of the Elderly Drivers' Health Status Scale, a2 is the score of the Elderly Drivers' Driving Ability Self-Assessment Scale, a3 is the score of the Elderly Drivers' Driving Skills Self-Assessment Scale, a4 is the score of the Elderly Drivers' Driving Ability Confidence Level Assessment Scale, and wi (i = 1, 2, 3, 4) is the weight for calculating the total score of the corresponding module.

[0029] Furthermore, in step 5, the cost function of road impedance is:

[0030] Then the cost of the road impedance in the driving path is f t It can be expressed as the weighted sum of four road impedances, namely:

[0031] f t =k1T1+k2T2+k3T3+k4T4

[0032] f t is the traffic impedance cost; T1 is the impedance caused by the intersection spacing, T2 is the road impedance caused by pedestrians crossing the road, T3 is the road impedance caused by non-motor vehicle interference, T4 is the road impedance caused by lane width change, and k is the weight coefficient of different impedances;

[0033] When the road traffic saturation is low, the Webster model is used to calculate the impedance, that is:

[0034]

[0035] When the saturation is high, the Akcelik model is used to calculate the impedance, that is:

[0036]

[0037] Where y is the saturation, calculated using the formula:

[0038] y=1-(1-k / k j ) 2

[0039]

[0040] In the above formula: c is the signal period, λ is the green-to-signal ratio, k is the traffic density, η i are different road traffic impedance coefficients, n is the number of motor vehicle lanes on the road, l0 is 1.5m, and l is 5m;

[0041] The unsignalized intersection is considered as a vehicle without traffic signal influence, which has the same driving characteristics as the highway. The improved BPR function model is used for calculation, namely:

[0042]

[0043] Among them, k j is the theoretical blocking density calculated by different influence coefficients, and α and β are the theoretical impedance influence coefficients.

[0044] Furthermore, step 6, safe travel path planning for elderly drivers; the safety target in the driving path can be calculated by the four types of costs in driving: safety reliability cost, driving confidence cost, driving ability cost, and road impedance cost. Characterize it, that is, the travel cost of elderly drivers is:

[0045]

[0046] Where k n is the weight of different cost functions. In this invention, the four types of costs are considered equally important. The values ​​of n are 1, 2, 3, and 4. For different travel costs;

[0047] The calculation principle of the A* (A-star) shortest path algorithm is to select a travel plan with higher travel cost and safety for elderly drivers; its core is to plan the path by minimizing the actual cost and estimated cost between two points. The algorithm will give priority to searching for nodes with high probability in the grid. In order to solve the high probability problem, define n s With n d are the initial node and the target node respectively, n i Indicates that any node is the current node, n b For the node with the smallest F, an evaluation function is introduced:

[0048] F(n)=G(n)+H(n)

[0049] Where G(n) represents the current node n i The minimum cost to the initial node is expressed as the shortest actual distance between two points in the shortest path; H(n) represents the current node n i To the target node n d The cost is expressed as the straight-line distance between two points in the shortest path;

[0050] The travel cost of elderly drivers is introduced into the estimation function, and the travel cost of elderly drivers at the node position is used as the vertical axis coordinate, and the coordinate position is the horizontal axis and vertical axis coordinate to construct the three-dimensional Cartesian coordinate, that is, the initial node n s The coordinates of (x s ,y s ,z s )Current node n i The coordinates are (x, y, z), and its target node n d The coordinates of (x d ,y d ,z d ), using the Euclidean distance formula in three-dimensional space, with the initial node n s Arrived at the current node n i The sum of the minimum geometric distances is used as the minimum cost G(n), and the geometric distance from the current node to the target node is used as the cost H(n):

[0051]

[0052] Beneficial effects of the present invention:

[0053] Since the driving behavior, psychological and physiological characteristics, and driving confidence of elderly drivers are taken into consideration, this path planning method is more in line with the actual needs of elderly drivers and can effectively improve their travel safety.

[0054] By constructing a travel cost function that includes multiple factors such as driving ability, driving confidence, safety reliability, and road impedance, the path planning results are made more objective and accurate.

[0055] This method can effectively find the path with the lowest travel cost by adopting the A* (A-star) shortest path algorithm, and realizes the rapid optimization of the driving route of elderly drivers.

[0056] This method can effectively judge the driving ability level of elderly drivers by evaluating their driving ability, help them reasonably choose travel routes, avoid overly complex or difficult driving environments, and improve travel safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1Flow chart of the method of the present invention;

[0058] Figure 2 The reliability and conflict complexity curve of the method of the present invention is shown in FIG.

[0059] Figure 3 This is a reliability change trend diagram of the method of the present invention;

[0060] Figure 4 is the potential category index table of the method of the present invention;

[0061] Figure 5 This is a comparison chart of the reliability of the method of the present invention in scenes with and without pedestrians;

[0062] Figure 6 It is the algorithm flow chart of the method of the present invention;

[0063] Figure 7 This is the initial road network diagram of the method of the present invention. DETAILED DESCRIPTION

[0064] The following drawings illustrate embodiments of the present application. For clarity, many practical details will be included in the following description. However, it should be understood that these practical details should not be construed as limiting the present application. In other words, these practical details are not essential to the embodiments described in this application. Furthermore, to simplify the drawings, some conventional structures and components are shown in simplified schematic form.

[0065] like Figure 1 As shown, the method includes

[0066] Step 1: Collect relevant data of elderly drivers;

[0067] In order to explore the safety and reliability of elderly drivers, it is necessary to obtain experimental data on their completion of various driving tasks in different scenarios. The experiment recruited elderly drivers with a certain driving age and experience, and without diseases that affect their driving ability. Through driving simulation experiments, the driving behavior characteristic parameters, psychological and physiological characteristics, and operational behavior data of the drivers were obtained. A questionnaire survey was also conducted on the elderly drivers' driving confidence in different driving scenarios. A total of 39 drivers were recruited for the experiment, including 22 young and middle-aged drivers.

[0068] There were 17 elderly drivers.

[0069] The experimental subjects of this invention are drivers of different ages. To ensure the validity of the experimental data, the recruited experimental personnel must have obtained a driver's license for more than one year and are still driving a motor vehicle recently. Since eye movement data and electrocardiogram data need to be collected, the drivers are required to be in good health, without eye diseases, heart diseases or other diseases that affect normal driving ability, and have completely independent perception, judgment and operation capabilities, and can complete various driving tasks normally.

[0070] Step 2: Constructing the cost function of safety and reliability of elderly drivers;

[0071] During the driving process, drivers are affected by multiple factors, including people, vehicles, roads, and the environment. The more complex the driving environment, the more conflicts there are during driving. The more complex the information and operations that drivers need to process in a short period of time, the greater the operational load on the driver, and the higher the possibility of danger to the driver. Considering that intersections are the most frequent accident sites and the most concentrated driving tasks on the road

[48] , from the perspective of drivers, especially elderly drivers, we explore the impact of different intersection conflicts on the driver subject and reflect the possibility of drivers safely completing driving tasks in different driving scenarios. This section will discuss the safety reliability of drivers and the risk of conflicts.

[0072] Driver reliability quantifies the probability of various errors, either external or internal, that occur during a driver's driving task, affecting the safe operation of the vehicle. This quantifies the driver's ability to safely complete the driving task, assessing the driver's ability to overcome various risks and complete the task based on operational and perception information. However, this requires access to the driver's perceptions and judgments during the task, which has certain limitations in practical applications. Conflict, a key factor in assessing risk at intersections, reflects the likelihood of a driver encountering danger during the driving task. The higher the conflict complexity, the greater the risk and challenge faced by the driver, and the greater the likelihood of failure. In other words, the higher the probability of conflict during the driving task, the greater the intersection conflict complexity, the higher the probability of driver error, and the lower the driver's safety reliability. Therefore, driver safety reliability at an intersection is considered the probability of a driver facing conflict risk at the intersection. The higher the conflict risk, the lower the driver's reliability. The cost function for driver safety reliability is:

[0073] f r =∑P i (1)

[0074] Among them, C1 represents the driver's safety reliability cost function, P i is the probability of each conflict risk in the intersection.

[0075] (1) SOR theory

[0076] The SOR theoretical model, namely the "stimulus (S) - organism (O) - response (R)" model, was first proposed by Mehrabian and Russell to improve the classic stimulus-response model. The model points out that various environmental factors can cause changes in an individual's psychology or emotions, thereby affecting the individual's behavior.

[63] , emphasizing the impact of the external environment on people. For drivers, the process of completing various driving tasks can be divided into three parts: information perception stage (S), information judgment stage (O), and driving operation stage (R).

[64] The information judgment stage is mainly for the brain to process and judge various types of road traffic information obtained through the sensory organs, make driving decisions and command the body to control the vehicle to ensure the safe operation of the vehicle.

[0077] Based on the SOR theoretical model, Japanese scholar Masakazu Iguchi regards human reliability as the joint result of information input reliability, judgment and decision reliability, and operation output reliability.

[65] ,Right now

[0078] γ=γ1·γ2·γ3 (2)

[0079] Then the reliability calculation of the driver in the driving task can be expressed as:

[0080] R(t)=R S (t)R O (t)R R (t) (3)

[0081] Where: R S (t) is the reliability of the driver’s perception characteristics in the driving task; R O (t) is the driver's judgment reliability in the driving task; R R (t) is the reliability of the driver's operating behavior in the driving task;

[0082]

[0083] Where: λ(t) is the error probability of the driver at different stages of the driving task;

[0084]

[0085] Where: μ is the mean of the corresponding indicator; σ is the standard deviation of the corresponding indicator.

[0086] (2) Intersection conflict complexity

[0087] Based on the research on intersection conflict points, merging conflict points, and diverging conflict points, a calculation formula for the average conflict probability of motor vehicles at different conflict points was constructed. Different conflict probabilities were assigned using reduction coefficients. The conflict complexity calculation of the intersection is as follows:

[0088]

[0089] Where: and are the average conflict probabilities before reduction at the diverging conflict point and the merging conflict point respectively; is the average conflict probability of the intersection conflict point; C is the complexity of motor vehicle conflict at the intersection; Q P is the main traffic flow, Q S For secondary traffic flow; N j 、N d 、N m They are the number of three types of conflict points: intersection, diversion, and confluence.

[0090] Constructing the cost function

[0091] (1) Reliability of Perception and Judgment Behavior

[0092] Young and middle-aged drivers differ from older drivers in terms of their visual characteristics, physiological and psychological characteristics, and operational behavior. For example, older drivers exhibit relatively higher average fixation duration and number of fixations across various scenarios. Considering that the magnitude of a glance reflects the driver's ability to acquire various types of traffic information, and the duration of a driver's gaze reflects the speed with which they process various types of information in the surrounding road and traffic environment, these two behaviors correspond to the driver's information perception and judgment and decision-making processes during driving tasks, respectively. Therefore, this section will use fixations and saccades to assess the reliability of drivers' perception and judgment behaviors.

[0093] First, taking the driver's glance amplitude data in scene 1 as an example, the driver's glance amplitude data is obtained by processing the eye tracker data, and the mean, standard deviation, and time in the scene are calculated, as shown in the table.

[0094] Table 1 Driver saccade amplitude data (part)

[0095]

[0096] Taking the calculation of the reliability of driver No. 1's glance behavior in scenario 1 as an example, substituting the various driver glance amplitude data in Table 4.1, using Formula 5 to calculate, the φ value of driver No. 1 is 618.5701, and Formula 6 can be used to calculate the error probability λ value of driver No. 1 to be 0.008475. Using Formula 4 to calculate, the reliability of driver No. 1's glance behavior is 0.991561. Repeating Formulas 4 to 6, the reliabilities of driver No. 1's glance behavior in the remaining five scenarios can be obtained as 0.991006, 0.987819, 0.985825, 0.986841, and 0.994821, respectively. The above calculations were performed on the glance behavior data of all drivers. Drivers were then divided into two groups according to age: young and middle-aged drivers and elderly drivers. The average values ​​were calculated for each age group to obtain the glance behavior reliability of drivers of different age groups. Similarly, the gaze behavior reliability was obtained. According to Equation 3, the two values ​​were multiplied together to obtain the perception judgment reliability of different driver groups. The calculation results are shown in Table 2.

[0097] Table 2 Calculation results of driver visual behavior reliability

[0098]

[0099] Comparing the reliability of gaze and scanning behavior between the two driver groups revealed that elderly drivers consistently performed at lower levels across various driver reliability indicators. This may be due to age-related visual decline (e.g., reduced horizontal and vertical visual fields), which in turn reduces their ability to perceive, judge, and process traffic information, leading to lower perception reliability. Furthermore, the results of visual behavior reliability calculations indicate that elderly drivers are more likely to make mistakes at intersections due to personal reasons, making intersections a greater influence on their safety and reliability.

[0100] (2) Operational behavior reliability

[0101] When a driver is driving a vehicle, he or she mainly controls the vehicle by turning the steering wheel and pressing the accelerator and brake pedals. Therefore, the changes in the accelerator, brake pedal, and steering wheel angle are also a direct reflection of the driver's driving behavior while driving the vehicle. They can well represent the reliability of the driver's operating behavior in different driving scenarios. Equations 4 to 6 are used to calculate the reliability of the driver's operating behavior in each scenario. The calculation results are shown in Table 3:

[0102] Table 3 Calculation results of driver operation behavior reliability

[0103]

[0104] Comparing the reliability of accelerator, pedal, and steering wheel operation behaviors of elderly drivers with those of young and middle-aged drivers reveals that elderly drivers have lower reliability results in all calculations, indicating that young and middle-aged drivers can perform various driving operations more accurately and sensitively when passing through intersections, react more quickly to changes in the traffic environment, and make timely adjustments to the vehicle's driving state. Secondly, comparing the reliability of the two types of drivers' operational behaviors, elderly drivers have lower reliability, lower than that of young and middle-aged drivers in all experimental scenarios. This suggests that when passing through intersections, elderly drivers may not be able to make timely adjustments to the vehicle's driving state in the face of changes in the traffic environment due to the deterioration of their physical functions and reduced muscle strength and flexibility, making them more prone to errors.

[0105] According to Formula 3, the driver reliability of young and middle-aged drivers and elderly drivers can be calculated. The calculation results are shown in Table 4 below:

[0106] Table 4 Calculation results of driver reliability

[0107]

[0108] A comparison found that the reliability of elderly drivers was lower than that of young and middle-aged drivers. The reasons for this were mainly due to the decline in physical functions and driving abilities of elderly drivers due to aging, especially the decline in visual ability, which affected their ability to perceive and judge various information in complex road traffic environments, resulting in lower reliability of their perception and judgment. Compared with young and middle-aged drivers, they were unable to detect risks in road traffic in a timely manner; secondly, due to physical decline, elderly drivers' driving skills and physical flexibility were restricted. Compared with young drivers, the reliability of their various operational behaviors was at a lower level, and their ability to control the vehicle was relatively weak. In an emergency, they were unable to make timely and effective adjustments to the vehicle's driving status; the three factors together led to a decrease in their driving reliability, and they were more likely to affect driving safety due to their own mistakes on the road.

[0109] (3) Intersection conflict complexity

[0110] The intersection conflict complexity under different scenarios can be obtained by calculating Equations 7 and 8. The calculation results are shown in Table 5:

[0111] Table 5 Calculation results of intersection conflict complexity

[0112]

[0113] According to the conflict complexity obtained in different scenarios and the reliability of the 39 drivers participating in the experiment, a reliability and conflict complexity change curve in different scenarios can be drawn, such as Figure 2It can be found that the reliability of elderly drivers has the same changing trend as that of young and middle-aged drivers, indicating that the impact of conflict complexity in the scene on drivers of different age groups is similar. With the scene conflict complexity as the horizontal axis, the changing trend of intersection conflict complexity and driver reliability in different scenes can be plotted, as shown in the figure below. Figure 3 :

[0114] According to the experimental results, at intersections, the trends of driver reliability and intersection conflict complexity differ. From the overall trend, as the complexity of the conflict increases, that is, the probability of crossing, merging, and diverging conflicts increases, the driver may face more risks during driving, and the difficulty of completing the driving task will increase, and the overall driver reliability will decrease accordingly. That is, when the complexity of the road intersection increases, the driver's driving reliability level will show a downward trend, and the decline in the reliability level of elderly drivers is more obvious than that of middle-aged and young drivers.

[0115] At the same time, considering that drivers' steering behaviors, such as turning the steering wheel, vary significantly when performing different steering tasks (left turn, right turn, and going straight), this may affect operational reliability. Therefore, we compared the changes in driver operational reliability when performing different steering tasks. As shown in the table, for young and middle-aged drivers and elderly drivers, the operational reliability of drivers in high-conflict complexity scenarios was lower than that in low-conflict complexity scenarios across different driving steering tasks, indicating that the impact of different steering tasks during driving is relatively small.

[0116] Table 6 Reliability comparison results

[0117]

[0118] The reliability level of the driver in different scenarios can be expressed as the weighted sum of the average conflict complexity under the three conflict forms. The cost function f considering the driver's safety reliability is: r (n) Such as:

[0119]

[0120] Where k is the weight coefficient, which is used to calculate the driving task according to the driving scenario; f r The larger (n) is, the greater the complexity of the intersection conflict is and the lower the driver reliability level is.

[0121] The weights assigned to different conflict probabilities are calculated using the severity of different conflict types. This method defines scores based on the severity of the accidents and losses caused by different conflict types. Here, the relative severity of different conflict types is calculated by rounding the severity of the accident rates and economic losses.

[0122] Table 7 Conflict point malignity

[0123]

[0124] k i =SC i / SC (10)

[0125] Formula 10 is used to calculate the weights of different types of conflict points under different driving tasks, where SC i The severity of different conflict points.

[0126] Taking intersection 1 as an example, according to formula 10, we can get k1 = 0.706, k2 = 0.118, k3 = 0.0.176. The safety reliability cost of the driver in scenario 1 is:

[0127]

[0128] In order to ensure that drivers can successfully complete their driving tasks, their reliability level on the road should be improved as much as possible and the influencing factor of intersection conflict should be reduced. The solution with the highest overall safety reliability level of drivers on the road can be expressed as:

[0129] F r =min∑f r (12)

[0130] Step 3: Constructing the cost function of elderly drivers’ driving confidence;

[0131] Since drivers have a certain degree of subjectivity in their questionnaire selection, directly using drivers' subjective driving ability evaluation to classify drivers may affect the results of driving scenario avoidance. Therefore, the latent class analysis (LCA) method will be used to classify drivers.

[0132] AIC (Akaike Information Criterion), BIC (Bayesian Information Criterion), aBIC (Akaike Bayesian Information Criterion) and Entropy (entropy) were used as evaluation criteria to evaluate the fit of the latent class model under different categories. The specific results are as follows Figure 1 As shown in the figure, when the model is divided into three categories, the AIC, BIC, and aBIC values ​​are all at a low level. Further increasing the number of categories results in very small changes in AIC and aBIC (around 1%), and the entropy value of this category is higher than 0.8, indicating a high accuracy of the results. Therefore, based on the balance between fitting, simplicity, and interpretability, the optimal number of latent classes is determined to be three.

[0133] The LCA classification results were analyzed using the multinomial logit model (MNL). The multinomial logit model uses the random utility or utility maximization assumption, is easy to solve, and has a wide range of applications. The table shows the significance (P < 0.05) results after using the multinomial logit model:

[0134] Table 8 Significance results

[0135]

[0136] Analysis of the significance results revealed similarities across different driving scenarios for the three categories of elderly drivers. All showed significant responses when driving during busy times and locations, at intersections without traffic signals, and at roundabouts, indicating a lack of confidence in their driving abilities in these scenarios. Furthermore, statistical analysis of the questionnaires, categorizing elderly drivers with below-average confidence scores as lacking confidence, revealed that 68% of elderly drivers lacked confidence in driving during busy times and locations, 64% at intersections without traffic signals, and 66% at roundabouts, potentially increasing the risk of driving. For safety reasons, elderly drivers may be more inclined to avoid these scenarios. In addition to these three scenarios, some elderly drivers also expressed a lack of confidence in other scenarios. For example, 16% of elderly drivers rated their confidence level at 1 on expressways. However, this percentage is relatively low compared to the other three scenarios, so they were grouped together in the calculation of lack of confidence. Assuming that the driver's confidence level in these scenarios is as follows:

[0137] f c =k1N1+k2N2+k3N3+k4N4 (13)

[0138] N i is the statistical number of different scenarios, k i is the weight coefficient of different scenarios; k is the weight coefficient of confidence loss in different scenarios; N1, N2, and N3 are the number of three low-confidence scenarios respectively; N4 is the number of scenarios other than the three low-confidence traffic scenarios, including driving scenarios such as expressways, and is counted according to the characteristics of intersections and road sections.

[0139] In order to further determine the impact of different scenario factors on the driving confidence of elderly drivers, it is necessary to assign weights to the lack of confidence in different scenarios. In the factor analysis method, the component score matrix is ​​generally used to characterize the relationship between different factors.

[72] , where the KMO test value is 0.82, greater than 0.600, and the Bartlett sphericity test probability is 0.000, less than 0.010, which meets the criteria for the use of factor analysis methods. After factor rotation of driving confidence, the component score matrix is ​​obtained. Through data standardization, it is found that the weight coefficient of driving at busy times and locations is 0.28, the weight coefficient of unsignaled intersections is 0.29, the weight coefficient of roundabouts is 0.32, and the weight coefficient of scenarios other than the three scenarios is 0.10. The driver confidence cost is:

[0140] f c =0.28N1+0.29N2+0.32N3+0.1N4 (14).

[0141] Step 4: Establishment of a subjective evaluation system for elderly drivers’ driving ability

[0142] The relevant scales for the self-assessment test of driving ability of the elderly include the Driver Basic Information Scale, the Driver Health Self-Report Scale, the Driving Ability Self-Assessment Scale (including four modules: driving situation, visual attention, decision-making ability, and execution ability), the Driving Skills Self-Assessment Scale, and the Driving Ability Confidence Level Scale, which can be used for self-assessment and self-testing of the driving ability of elderly drivers. Among them:

[0143] a. Driver Basic Information Scale

[0144] The Driver Basic Information Scale is used to collect statistics on the test subjects' gender, age, education level, employment status, driving experience, driving mileage, etc.; it also includes the test subjects' relevant driving exposure information, including average weekly driving frequency, average weekly driving mileage, number of traffic violations in a year, number of traffic accidents in a year, and evaluation of their own psychological state while driving.

[0145] b. Driver's Health Self-Report Scale

[0146] The Driver Health Self-Report Scale mainly investigates elderly drivers' overall evaluation of their health status, medical history, medication use, drinking habits, driving interruption due to health problems, and driving status.

[0147] c. Driving Ability Self-Assessment Scale

[0148] The Driving Ability Self-Assessment Scale consists of four modules: driving conditions, visual attention, decision-making ability, and executive ability. The driving conditions module examines the driver's experience on unfamiliar roads, driving at night, inclement weather, congested roads, driving conditions, and driving difficulties. Visual attention modules examine the driver's vision, concentration difficulties, visual neglect, and visual misjudgment. Decision-making ability modules examine the driver's multi-tasking driving conditions, concentration, comprehension, and memory. Executive ability modules examine the driver's physical condition while driving, their ability to execute driving operations, and their ability to respond to emergencies.

[0149] d. Driving Skills Self-Assessment Scale

[0150] The Driving Skills Self-Assessment Scale mainly investigates drivers' self-assessment of driving tasks such as performing safety checks, using appropriate braking methods at the right time, having a good sense of direction in various traffic environments, interacting with other road users, driving economically, keeping a sufficient safe distance while driving, adjusting speed according to real-time driving conditions, abiding by traffic rules, anticipating the development of different emergencies in traffic, identifying risks in various traffic conditions, and accurately operating the car.

[0151] e. Driving Ability Confidence Level Scale

[0152] The Driving Ability Confidence Level Scale mainly investigates the driver's confidence level in driving at night, driving in bad weather, driving at night in bad weather, driving on highways, driving in unfamiliar places, driving at busy traffic times or places, driving at intersections without traffic signals, driving at roundabouts, driving at times or places with many non-motor vehicles, driving alone, and driving long distances or for long periods of time.

[0153] Constructing the cost function

[0154] A comprehensive evaluation of elderly drivers' driving ability is conducted based on their self-assessment scores on the health status, driving ability, driving skills, and confidence level scales. Different weights are assigned to each module in the total score calculation based on the number of questions and corresponding scores in the designed scale. The calculation method is as follows:

[0155]

[0156] Where wi is the weight of the ith submodule, ti is the score of the ith submodule, i = 1, 2, 3, 4, and ttotal is the total score of the test scale.

[0157] The formula for calculating the total score of the subjective assessment of driving ability is as follows:

[0158]

[0159] Where sj is the total score of the subjective evaluation of driving ability in the jth test, and the total score range is [0,100]. Sij is the score of the i-th submodule in the j-th test, where i = 1, 2, 3, 4, j∈N*. Ttotal is the full score of the test scale.

[0160] The total score of the elderly driver's subjective driving ability assessment can be used to determine the overall driving ability evaluation level. The total score ranges from 0 to 100, with each 10-point interval representing a level, corresponding to (0-60) poor, (60-70) fair, (70-80) good, (80-90) excellent, and (90-100) excellent. The test results categorize the subject's driving ability into different levels, which can influence the content of subsequent intervention training. Furthermore, the driving ability cost function for elderly drivers can be expressed as the weighted sum of the scores of four modules: health status, driving ability, driving skills, and driving confidence, namely:

[0161] f a =w1a1+w2a2+w3a3+w4a4(17)

[0162] where a1 is the score of the Elderly Drivers' Health Status Scale, a2 is the score of the Elderly Drivers' Driving Ability Self-Assessment Scale, a3 is the score of the Elderly Drivers' Driving Skills Self-Assessment Scale, a4 is the score of the Elderly Drivers' Driving Ability Confidence Level Assessment Scale, and wi (i = 1, 2, 3, 4) is the weight for calculating the total score of the corresponding module.

[0163] Step 5: Construct the cost function of road impedance;

[0164] The main considerations are the distance between adjacent intersections, pedestrian crossings, non-motor vehicles, and lane width. Furthermore, the impact of different impedance factors on driving safety in the subjective perception of elderly drivers is comprehensively considered. The goal is to select roads with lower impedance costs, smoother driving, and higher comfort for elderly drivers. Therefore, the four road impedances present in the road are considered as part of the travel cost. The higher the impedance, the higher the corresponding travel cost. The cost of road impedance in the driving path, f t It can be expressed as the weighted sum of four road impedances, namely:

[0165] f t =k1T1+k2T2+k3T3+k4T4 (18)

[0166] f t is the traffic impedance cost; T1 is the impedance caused by the intersection spacing, T2 is the road impedance caused by pedestrians crossing the road, T3 is the road impedance caused by non-motor vehicle interference, T4 is the road impedance caused by lane width change, and k is the weight coefficient of different impedances;

[0167] When the road traffic saturation is low, the Webster model is used to calculate the impedance, that is:

[0168]

[0169] When the saturation is high, the Akcelik model is used to calculate the impedance, that is:

[0170]

[0171] Where y is the saturation, which is calculated using formula (20):

[0172] y=1-(1-k / k j ) 2 (twenty one)

[0173]

[0174] In the above formula: c is the signal period, λ is the green-to-signal ratio, k is the traffic density, η i are different road traffic impedance coefficients, n is the number of motor vehicle lanes on the road, l0 is 1.5m, and l is 5m.

[0175] The unsignalized intersection is considered as a vehicle without traffic signal influence, which has the same driving characteristics as the highway. The improved BPR function model is used for calculation, namely:

[0176]

[0177] Among them, k j is the theoretical blocking density calculated by different influence coefficients, α and β are the theoretical impedance influence coefficients, and the recommended values ​​are 0.15 and 4.

[0178] Determination of road impedance influence coefficient

[0179] Intersection spacing influence coefficient η1

[0180] Existing research has shown that in addition to the control method of road intersections, the distance between two road intersections on the road also has a significant impact on road capacity. If the distance between two adjacent intersections on the road is too short, the vehicle delay time will account for a higher proportion of the overall time, which is not conducive to vehicle driving. Comparing the intersection spacing in the experiment with the reliability of elderly drivers, it was found that when the intersection spacing was reduced from 1062m to 428m, the average reliability of elderly drivers decreased from 0.78 to 0.68, indicating that the intersection spacing has an impact on the safety and reliability of elderly drivers in completing driving tasks, and this should be considered in the road impedance cost. Among them, the calculation formula for the intersection impact correction coefficient is as follows:

[0181]

[0182] (2) Pedestrian interference influence coefficient η2

[0183] In order to verify whether pedestrian crossing has a substantial impact on elderly drivers, we selected intersections with pedestrians and intersections without pedestrians for comparison, and calculated the operational reliability of elderly drivers in different scenarios. The operational reliability of some elderly drivers is as follows: Figure 5 .

[0184] Taking Scene 1 without pedestrians and Scene 2 with pedestrians as examples, it can be found that the operating reliability of most elderly drivers in the scene with pedestrians is lower than that in the scene without pedestrians, and the impact is greater for some elderly drivers. For example, the driving operation reliability of elderly driver No. 1 decreased by nearly 10% when pedestrians were crossing the road. The change is quite obvious. Pedestrians crossing the road has a real impact on elderly drivers.

[0185] Due to the high uncertainty of pedestrian movement, their walking speed, crossing time, and traffic volume are all highly random. In particular, the timing of their crossings is difficult to predict. Therefore, pedestrian crossings are divided into two situations. First, on roads without facilities such as underpasses, crosswalk lights, overpasses, or central medians, the time and location of pedestrian crossings are uncertain, significantly impacting vehicles normally traveling on the road. The pedestrian interference coefficient is determined based on the land area and the degree of pedestrian interference with traffic flow: 0.5-0.6 for busy commercial areas, 0.7-0.8 for general commercial areas, and 0.9-1.0 for general roads. Second, on roads equipped with crosswalk lights and underpasses, pedestrians cannot normally cross the road. It is assumed that pedestrians follow traffic rules and do not directly conflict with normally traveling vehicles, resulting in zero impact on normally traveling vehicles. Therefore, on roads with safety facilities, the pedestrian interference coefficient is set to 1.

[0186] (3) Non-motor vehicle impact coefficient η3

[0187] In most cities today, a separation strip is designed between motor vehicle lanes and non-motor vehicle lanes at the beginning of construction, and the road signal control facilities are relatively complete. During normal driving, there is less mixing of motor vehicles and non-motor vehicle lanes, and the impact of non-motor vehicles on motor vehicles is also relatively small. Therefore, when a separation strip is set between non-motor vehicles and motor vehicles, it is considered that non-motor vehicles have no impact on motor vehicles under normal driving conditions, and the value is 1. When there is no separation strip design, the impact of non-motor vehicles on motor vehicles can be determined based on the degree of non-motor vehicle encroachment on motor vehicle lanes. 77] To obtain:

[0188]

[0189] Q a , Qb is the traffic volume of the non-motorized vehicle lane and the actual traffic capacity of the non-motorized vehicle lane; W1 and W2 are the lane widths of the motor vehicle lane and the non-motorized vehicle lane.

[0190] (4) Lane width influence coefficient η4

[0191] When the width of a motor vehicle lane on urban roads in my country is less than the standard lane width of 3.5m, the vehicle's speed will be affected. When the lane width is greater than the standard width of 3.5m, the driver has more ample driving space, and the speed will gradually increase until it reaches the limit of the driver or vehicle. 77] :

[0192]

[0193] After calculating the road impedance coefficients in four aspects, substituting them into equations 16 to 20 can obtain the corresponding impedances T1, T2, T3, and T4.

[0194] Determination of cost function weight coefficient

[0195] Considering that elderly drivers may have varying impacts on driving safety due to the decline in their physical function and driving ability, a questionnaire was designed to investigate the impact of road impedance factors on elderly drivers' driving safety. Specifically, the questionnaire asked drivers to rate the impact of four factors, namely, intersection spacing, pedestrians, lane width, and non-motor vehicles, on their driving safety. After screening, a total of 85 valid questionnaires were obtained, and the average scores for each factor are shown in the table below:

[0196] Table 9 Average score table

[0197]

[0198] By comparison, we can find that elderly drivers score higher on all factors, indicating that they believe that various impedance factors have a greater impact on their driving safety and are more likely to affect their safe driving. At the same time, the degree of influence of different factors also varies. For young and middle-aged drivers, the lowest score, that is, the factor with the least impact on driving safety, is pedestrians, while for elderly drivers, it is non-motor vehicles. In summary, the same impedance factor has different effects on drivers of different ages. In order to clarify the differences in the impact of different factors when calculating the cost of road impedance, the entropy weight method is used to assign weights to different factors. The calculation results are shown in the table:

[0199] Table 10 Calculation results of entropy weight method

[0200]

[0201] Therefore, for elderly drivers, the weights of the four factors of intersection spacing, pedestrians, lane width, and non-motor vehicles are W = (0.14, 0.19, 0.28, 0.37). The cost function considering the road impedance factor is:

[0202] f t =0.14T1+0.19T2+0.28T3+0.37T4 (27)

[0203] Step 5: Plan safe travel routes for elderly drivers.

[0204] In summary, the safety target in the driving path can be achieved through the four types of costs: safety reliability cost, driving confidence cost, driving ability cost, and road impedance cost. Characterize it, that is, the travel cost of elderly drivers is:

[0205]

[0206] Where k n is the weight of different cost functions, the value of n is 1, 2, 3, For different travel costs.

[0207] The main purpose of optimal path selection is to use the reliability level of different intersections, the driving confidence of elderly drivers in different scenarios, and the impedance of the road as the travel cost, and the level of risk cost as the core of the evaluation of travel path selection. Based on the calculation principle of the A* (A-star) shortest path algorithm, the A* algorithm is a heuristic intelligent algorithm. Its core is to plan the path by minimizing the actual cost and estimated cost between two points. The algorithm will prioritize the nodes with high probability in the grid. In order to solve the high probability problem, it defines n s With n d are the initial node and the target node respectively, n i Indicates that any node is the current node, n b For the node with the smallest F, an evaluation function is introduced:

[0208] F(n)=G(n)+H(n) (29)

[0209] Where G(n) represents the current node n i The minimum cost to the initial node is expressed as the shortest actual distance between two points in the shortest path; H(n) represents the current node n i To the target node n d The cost is expressed as the straight-line distance between two points in the shortest path.

[0210] The present invention takes the highest travel safety as the research point, introduces the travel cost of elderly drivers into the estimation function, takes the travel cost of elderly drivers at the node position as the vertical axis coordinate, and the coordinate position as the horizontal axis and vertical axis coordinate to construct the three-dimensional Cartesian coordinate, that is, the initial node n s The coordinates of (x s ,y s ,z s )Current node n i The coordinates are (x, y, z), and its target node n d The coordinates of (x d ,y d ,z d ), using the Euclidean distance formula in three-dimensional space, with the initial node n s Arrived at the current node n i The sum of the minimum geometric distances is used as the minimum cost G(n), and the geometric distance from the current node to the target node is used as the cost H(n):

[0211]

[0212] The optimal path selection process is as follows Figure 6 As shown:

[0213] Step 1: Build a road network and name different intersections.

[0214] Step 2: Taking the first intersection G from the starting point as an example, set the starting point as the origin of the road network coordinate, and its coordinates are (0,0,0,).

[0215] Step 3: Count the traffic volume in scenario G and use Equations 9 and 10 to calculate the safety reliability cost f of the elderly driver in scenario G. r(G) is 0.12074; intersection G is a two-phase signal-controlled intersection, and the driver's confidence cost f c is 0.1; at the same time, the road width is 3.5m, with a complete motor vehicle and non-motor vehicle separation zone, and there is a pedestrian crossing light at the intersection. The impact of pedestrians on motor vehicles is small, so the non-motor vehicle impact coefficient is 1, and the pedestrian impact coefficient is 1. The impedance cost f from the starting point to the intersection G is calculated. t is 0.01073; then the cost C from the starting point to the intersection G G is 0.23147, then the coordinates of intersection G can be obtained as (0, 0.428, 0.23147). Using formula 26, the cost H(G) from the starting point to intersection G can be calculated as 0.4865823. Repeat the steps to obtain the coordinates of all nodes in the road network.

[0216] Step 4: Obtain the safest path: starting point → G → H → L → K → J → end point, and use this path as the best path B.

[0217] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A safe travel route planning method for elderly drivers that takes travel costs into consideration, characterized by: The method includes: Step 1: Collect relevant data of elderly drivers; Step 2: Construct a cost function for the safety and reliability of elderly drivers; The reliability level of the driver in different scenarios is expressed as the weighted sum of the average conflict complexity under three conflict forms: intersection conflict, divergence conflict, and merging conflict. The cost function f considering the driver's safety reliability is: r (n) Such as: Where k is the weight coefficient, which is calculated for the driving task according to the driving scenario; is the average conflict probability of the intersection conflict points; is the average conflict probability before reduction at the diversion conflict point; is the average conflict probability before reduction at the converging conflict point; N j is the number of cross-conflict points; N d is the number of diversion conflict points; N m is the number of confluence conflict points; Step 3: Construct a cost function for elderly drivers’ driving confidence; Among them, the lack of driving confidence on the road is obtained as: f c =k1N1+k2N2+k3N3+k4N4; N i is the statistical number of different scenarios, k i is the weight coefficient of different scenarios; k is the weight coefficient of confidence loss of different scenarios; N1, N2, N3 are the number of three low-confidence scenarios, and N4 is the number of scenarios other than the three low-confidence traffic scenarios; Step 4: Construct a cost function for the driving ability of elderly drivers; Among them, the driving ability of elderly drivers is comprehensively evaluated based on their self-assessment scores on the health status, driving ability, driving skills and confidence level scales; Step 5: Construct the cost function of road impedance; The cost of road impedance in the driving path f t It represents the weighted sum of four road impedances, namely: f t =k1T1+k2T2+k3T3+k4T4; f t is the weighted sum of four road impedances: intersection spacing, pedestrian crossing, non-motorized vehicles, and lane width; T1 is the impedance caused by intersection spacing, T2 is the road impedance caused by pedestrian crossing, T3 is the road impedance caused by non-motorized vehicle interference, and T4 is the road impedance caused by lane width change. k is the weight coefficient of different impedances. Step 6: Planning safe travel routes for elderly drivers; the safety target in the driving route is calculated by the four types of costs: safety reliability cost, driving confidence cost, driving ability cost, and road impedance cost. Characterize it, that is, the travel cost of elderly drivers is: Where k n is the weight of different cost functions. In this invention, the four types of costs are considered equally important. The values ​​of n are 1, 2, 3, and 4. The cost of safety and reliability, driving confidence, driving ability, and road impedance is defined. Based on the calculation principle of the A* (A-star) shortest path algorithm, a travel plan with higher travel costs, i.e., a safer travel plan, is selected for elderly drivers. Its core is to plan a path by minimizing the actual cost and estimated cost between two points. The algorithm prioritizes nodes with high probability in the grid. To address high-probability issues, ns and nd are defined as the initial and target nodes, respectively. ni represents any node, i.e., the current node. nb represents the node with the smallest F. An evaluation function is introduced: F(n)=G(n)+H(n); In the formula, G(n) represents the minimum cost between the current node ni and the initial node, which is expressed as the shortest actual distance between two points in the shortest path; H(n) represents the current node n i To the target node n d The cost is expressed as the straight-line distance between two points in the shortest path; The travel cost of elderly drivers is introduced into the estimation function. The travel cost of elderly drivers at the node position is used as the vertical axis coordinate, and the coordinate position is used as the horizontal and vertical axis coordinates to construct a three-dimensional Cartesian coordinate. That is, the coordinates of the initial node ns are (xs, ys, zs), the coordinates of the current node ni are (x, y, z), and the coordinates of its target node nd are (xd, yd, zd). Using the Euclidean distance formula in three-dimensional space, the sum of the minimum geometric distances from the initial node ns to the current node ni is used as the minimum cost G(n), and the geometric distance from the current node to the target node is used as the cost H(n):

2. The method for planning safe travel routes for elderly drivers considering travel costs according to claim 1, characterized in that: The step 1, collecting relevant data of elderly drivers, recruits elderly drivers with a certain driving age and experience and no diseases that affect driving ability, obtains the drivers' driving behavior characteristic parameters, psychological and physiological characteristics, and operational behavior data through driving simulation experiments, and conducts a questionnaire survey on the elderly drivers' driving confidence in different driving scenarios.

3. The method for planning safe travel routes for elderly drivers considering travel costs according to claim 1, characterized in that: In step 4, different total score calculation weights are assigned to each module according to the number of questions and corresponding scores of each sub-module of the designed scale. The calculation method is shown in the following formula: Where wi is the weight of the ith submodule, ti is the score of the ith submodule, i = 1, 2, 3, 4, and ttotal is the total score of the test scale; The formula for calculating the total score of the subjective evaluation of driving ability is as follows: Where sj is the total score of the subjective evaluation of driving ability in the jth test, and the total score range is [0,100]; sij is the score of the i-th submodule in the j-th test, i = 1, 2, 3, 4, j∈N*; ttotal is the total score of the test scale; The total score of the subjective evaluation of the elderly driver's driving ability is used to obtain the comprehensive evaluation result of their driving ability. The total score range is between 0 and 100 points, and each 10-point interval within the range is a level assessment, corresponding to (0-60) poor, (60-70) medium, (70-80) good, (80-90) excellent, and (90-100) excellent. The test results divide the subject's driving ability into different levels; The driving ability cost function of elderly drivers is expressed as the weighted sum of the scores of the four modules: health status, driving ability, driving skills, and driving ability confidence level, that is: <h2 style=";text-align:left;direction:ltr">f<h2 style=";text-align:left;direction:ltr"> a <h2 style=";text-align:left;direction:ltr"> (w1a1+w2a2+w3a3+w4a4) where a1 is the score of the Elderly Drivers' Health Status Scale, a2 is the score of the Elderly Drivers' Driving Ability Self-Assessment Scale, a3 is the score of the Elderly Drivers' Driving Skills Self-Assessment Scale, a4 is the score of the Elderly Drivers' Driving Ability Confidence Level Assessment Scale, and wi (i = 1, 2, 3, 4) is the weight for calculating the total score of the corresponding module.

4. The method for planning safe travel routes for elderly drivers considering travel costs according to claim 1, characterized in that: In the step 5, When the road traffic saturation is low, the Webster model is used to calculate the impedance, that is: When the saturation is high, the Akcelik model is used to calculate the impedance, that is: Where y is the saturation, calculated using the formula: y=1-(1-k / k j ) 2 ; In the above formula: c is the signal period, λ is the green-to-signal ratio, k is the traffic density, η i are different road traffic impedance coefficients, n is the number of motor vehicle lanes on the road, l0 is 1.5m, and l is 5m; The unsignalized intersection is considered as a vehicle without traffic signal influence, which has the same driving characteristics as the highway. The improved BPR function model is used for calculation, namely: Where kj is the theoretical blocking density calculated by different influence coefficients, and α and β are the theoretical impedance influence coefficients.

Images