| Autonomous driving that is based on the intelligent vehicle carries the vision of "zero casualties" in transportation and represents the strategic commanding heights of future automotive technology.However,limited by the law of technological development,laws and regulations,and ethical issues,it will still be a fact that the drivers are participants or even driving subjects in the driving process for a long time.With the continuous development of intelligent vehicle technology,the relationship between the vehicle and the driver has become extremely complex.Various machine intelligent control systems based on multi-source information perception and drivers with different behaviors will jointly constitute a humanmachine parallel control system for intelligent vehicle.The dynamic interaction between the intelligent system and the human driver may restrict each other in different driving environments and working conditions,or even results in conflicts,which will not only seriously affect the driving experience of the driver and the performance of the intelligent system,but also may pose a serious potential threat to driving safety.In view of the current situation that the level of vehicle intelligence cannot quickly meet the requirements of autonomous driving and the human-machine interaction relationship is increasingly complex and changeable,the concept of human-machine shared driving has received extensive attention and research,and has become the only way which must be passed until the fully automated driving come true.This paper,which is relies on the 13 th Five-Year National Key R&D Program Project " Human-Machine Shared driving Interaction Theory for Intelligent Electric Vehicle"(2016YFB0100904)and the National Natural Science Foundation of China Project "Research on Human-machine Parallel Control Conflict Mechanism and Key Technology of Cooperative Co-piloting for Intelligent Vehicles"(51775235),researches and establishes a human-machine haptic interactive shared steering control method based on driver behavior understanding for intelligent vehicle,which takes haptic interaction as the humanmachine interaction mode and aims at steering control of the vehicle.The focus is on coordinating the control roles of driver and intelligent system at the control execution layer,building a bridge for human-machine haptic interaction,establishing a channel for humanmachine intelligence integration,and achieving harmonious human-machine interaction and conflict resolution.The purpose is to improve vehicle safety and reduce the driving load of the driver.This paper revolves around two key scientific problems: the extraction and understanding of driver behavior under the state of human-machine torque coupling,and the mechanism of the driving authority allocation which reflects human-machine haptic interaction and conflict resolution.Therefore,this paper proposes risk awareness based on eye movement behavior understanding and steering initiative based on grip strength behavior understanding to form the understanding of driver behavior from two aspects of psychological implicit state and physical explicit state respectively,providing guidance and basis for the allocation of driving authority.On this basis,a haptic interactive shared steering control method based on variable virtual impedance is proposed to provide a way for continuous and dynamic allocation of driving authority.The forward and backward combination impedance coordination strategy is further constructed to form the driving authority allocation mechanism based on the understanding of the driver behavior to achieve harmonious human-machine interaction and conflict resolution.Firstly,the construction method of spatio-temporal risk field based on artificial potential field is studied.By constructing spatio-temporal risk field,the distribution of dynamic traffic risk in specific time span and road section is depicted in time-space domain,providing basis for the driver risk awareness identification.Traffic elements are divided into concrete elements and abstract elements according to their nature and influence mechanism.A bias sweeping method is developed for concrete elements to generate their risk distribution,and a risk distribution generation method based on Gaussian model is developed for abstract elements.Through the example scenario analysis,it is proved that the spatio-temporal risk field can describe the distribution of dynamic traffic risk in space-time domain in a unified and intuitive way.Secondly,the evaluation method of driver risk awareness based on visual characteristics is studied.In order to describe the characteristics of visual cognition of the driver,a visual perceive model based on Gamma distribution is established.Treating the eye movement behavior as the system incentive and the changing process of perceive level described by Gamma distribution as the system response,the perceive level of the driver is calculated by using convolution.The real vehicle field experiment for acquiring eye movement behavior is carried out,and the visual perceive model is calibrated based on the real vehicle data.Taking the spatio-temporal risk field as the objective distribution of traffic risk and the visual perceive model as the subjective perceive level of driver,the risk awareness of the driver is calculated by combining the two parts.An example is given to prove that the method achieves the quantitative estimation of driver risk awareness.Thirdly,the identification method of driver steering initiative based on grip force distribution is studied.An intelligent steering wheel is fabricated by arranging flexible film pressure sensor on the steering wheel and the data acquisition experiment of grip force distribution is carried out.Taking the steering impedance as the parameterized description of the neuromuscular system of the upper limb of the driver,the characteristic parameters of the steering impedance are identified based on the experimental data,and a method for the representation of steering initiative is established based on the steering impedance.A cascaded model composed of convolutional neural network and gated recurrent unit is established,and on this basis the steering initiative is calculated according to the representation relationship between steering impedance and steering activity.The validity of the identification method is verified by test.Fourthly,the haptic interactive shared steering control method based on variable virtual impedance is studied.A haptic interaction framework based on impedance concept is proposed.Based on this framework,a steering control algorithm based on variable virtual impedance is developed,and the influence of virtual impedance on driver hand feeling and control system performance is analyzed,which provides a way for driving authority allocation.The forward and backward combined impedance coordination strategy is constructed to coordinate the virtual impedance of the intelligent system with the driver behavior state,and a driving authority allocation mechanism based on the driver behavior understanding is formed.The simulation results show that this method describes the process of human-machine haptic interaction and can provide an effective way for the allocation of driving authority.Finally,the proposed method is tested based on human-machine parallel in-the-loop test.The human-machine parallel in-the-loop test platform is built,and six kinds of tests and three impedance coordination modes are designed for comparison.The experimental results show that the proposed method can make forward presetting and backward correction of virtual impedance according to the understanding of driver behavior,which forms a good allocation of driving authority through impedance coordination and achieves harmonious humanmachine haptic interaction and conflict elimination.Through objective evaluation and analysis,it is further verified that the established method can form a positive benefit to drivers and provide drivers with a safer and more comfortable driving experience. |
PDF Full Text Request