| Autonomous driving cars are a hot research spot at present,and more and more autonomous driving cars have begun to enter the actual road test stage.In practical terms,autonomous driving cars provide a ride service,.however,there are a large proportion of motion sickness in Chinese automobile passengers,that is,people who are easy to get motion sickness.Practical experience shows that there is a close relationship between vehicle motion state and motion sickness.Research suggests that autonomous driving cars could make car sickness more pronounced.Therefore,if the occupant comfort characteristics are considered in the control algorithm of autonomous driving vehicles,the travel service quality of autonomous driving vehicles can be effectively improved and the acceptance of autonomous driving vehicles by the public can be promoted.Relevant researches in the field of automobile dynamics show that vehicle movement is the direct factor that leads to the change of vehicle comfort,so the research on the relationship between the comfort of autonomous driving vehicles and vehicle movement is the key to solve the problem of vehicle comfort.To explore autonomous driving vehicles motion characteristics and the relationship between the occupant comfort,based on the actual driving by experiments,simulate the typical motion state of unmanned vehicles running state,under the motion state of the same comfort analysis of test subjects,determine the motion characteristics of the occupant comfort,for the unmanned vehicle control algorithm considers occupant comfort to provide technical support.The main research contents are as follows:(1)A total of 36 test subjects were recruited in a voluntary and paid way to carry out the real ride comfort evaluation experiment in the car test ground of Chang ’an University.According to individual physiological characteristics,passengers were classified from three aspects: gender,motion sickness susceptibility and motion sickness status.During the experiment,Volkswagen lavatory was selected as the test vehicle,and the vehicle acceleration,deceleration and turning operating were designed and implemented to simulate the real running state of autonomous driving cars.During the test,the occupant of the subject is not clear about the motion state of the vehicle.In the test,the driver’s head indicates the beginning and end of the occupant’s operation.After the end of a single operation,the occupant should give a real-time evaluation of the current ride experience to simulate the real ride condition.During the test,the evaluation result acquisition device and IOS intelligent device were used to collect the subjective comfort evaluation results of the test subjects on the vehicle acceleration,deceleration and turning operation as well as the relevant motion state parameters of the vehicle in real time,and a total of 1296 groups of experimental data were obtained.(2)The statistical analysis of the vehicles in the acceleration,deceleration and the turning of the single operation time,acceleration and jerk,and obtained the complete vehicle acceleration,deceleration and normal operation of turning,the single operation duration,in turn,is about 8s ～ 14 s,4s ～ 8s,6s ～ 18 s,acceleration range,in turn,is about 0.06 g ～ 0.15 g,0.1g ～ 0.5g and 0.1g ～ 0.4g,and jerk in turn about range is 0(g/s)～ 0.2(g/s),0.2(g/s)～ 0.8(g/s),0.1(g/s)～ 0.6(g/s).(3)The single factor analysis method was used to determine the factors affecting the passenger evaluation results,and the binary Logistic regression analysis model was introduced to determine the specific factors and influence effects: Under accelerated conditions,passengers prone to carsickness were more likely to be evaluated as uncomfortable than those not prone to carsickness.The greater the longitudinal acceleration of the vehicle,the more likely the passenger would be uncomfortable.Under deceleration condition,passengers prone to motion sickness are more likely to be evaluated as uncomfortable than passengers not prone to motion sickness.The larger the longitudinal acceleration of the vehicle is,the more likely it is to cause the discomfort of the passenger.Moreover,the acceptable longitudinal acceleration of the vehicle for passengers prone to motion sickness is significantly lower than that for passengers not prone to motion sickness.Under turning condition,female occupant is more likely to be judged as uncomfortable than male occupant.The greater the lateral acceleration of the vehicle,the more likely the occupant will be uncomfortable.Therefore,the longitudinal and lateral acceleration of a vehicle are the key factors affecting the ride comfort of the occupant,and there are significant differences in the ride comfort experience of occupant with different gender and motion sickness susceptibility.The above conclusions can provide ideas and materials for the establishment of driving modes with different occupant considerations for driverless vehicles.(4)Aiming at the analysis of occupant comfort characteristics studied in this paper,the occupant comfort prediction model is established based on the Bi-directional LSTM network model in Python software.The basic input variables of the model were divided into four input parameter combinations according to vehicle motion state parameters and individual physiological characteristics parameters,and the prediction accuracy of the model under different input parameter combinations was discussed.The analysis results show that the accuracy of the prediction model is 75% when the longitudinal acceleration,lateral acceleration and motion state of the vehicle are considered.Considering vehicle longitudinal and lateral acceleration and motion state classification,the accuracy of the model is 69%.Considering the acceleration,jerk and motion state classification,the accuracy of the prediction model is 79%.After taking into account the variables of vehicle acceleration,jerk,motion state classification,occupant gender and motion sickness susceptibility,the accuracy of the prediction model reached a maximum of 84%.The above variation rules of accuracy indicate that the correlation between vehicle acceleration and comfort is higher than that between acceleration,and the addition of individual characteristics of passengers is also of great significance for improving the accuracy of the model.(5)The research results of this paper shows that in the control algorithm system of unmanned vehicle,the control law of horizontal and longitudinal acceleration and jerk should be mainly considered to ensure that the values do not fall into the uncomfortable zone.At the same time,there is no car control algorithm should also through various technical means,occupant’s individual characteristics,and based on the individual characteristics of crew real-time adjustment of unmanned vehicle control algorithm,avoid no one occupant is in uncomfortable position,inside the car can theoretically achieve personalized passengers custom algorithm for unmanned vehicle,in order to promote the application of the unmanned vehicle and receive degrees. |
