Research On Evaluation Method Of Control Authority Transition In Intelligent Vehicle Considering Human-Vehicle-Road Characteristics

Intelligent vehicle can completely replace the driver for vehicle control under certain conditions while assisting the driver to complete the driving task.Intelligent vehicle provides drivers with different ways of maneuvering cars and also brings various human-machine interaction problems.Control authority transitions are initiated by the driver or the intelligent driving system,then the dynamic driving tasks are resumed by either of them,which have an important impact on the driving safety of intelligent vehicle.In this paper,considering the traffic environment characteristics and constraints,the driver's musculoskeletal characteristics and the vehicle moving state,a method of evaluating the control authority transitions was proposed.The main research contents were as follows.Firstly,the radar sensor was used for multi-target tracking and classification in the real vehicle environment to realize the dynamic traffic situation assessment and feature extraction including target-lane position,road curvature,guardrail information,moving targets density and etc.The driver's take-over behavior test in different traffic environments showed that the complexity of the traffic environment significantly affected the take-over time and quality.In complex traffic environments,the driver needed longer reaction time to recover sufficient situation awareness,resulting in severer maneuver.The moving target density had a significant impact on the complexity of the traffic environment,but the vehicle's trajectory curvature and the adjacent lane target state not.In order to identify the complexity of the traffic environment,this paper adopted a classifier based on the generalized linear model(GLM)to achieve accurate identification.Considering control authority transitions,this paper proposed the traffic environment constraints for safe take-over under typical conditions.At the same time,according to the characteristics of traffic environment,the evaluation criteria of control authority transition were established from the aspects of benefit,environment constraints and feasibility.Secondly,based on the principle of human biomechanics,the driver's musculoskeletal model was established and typical driving behavior simulation was carried out.It was confirmed that the main activating muscles during steering maneuver included the deltoid scapula,the infraspinatus muscle,the latissimus dorsi,the subscapularis muscle,the triceps lateral head and the triceps long head while braking mainly recruited tibialis anterior,the gastrocnemius,the rectus femoris,and the hamstring.A linear regression analysis between muscle activity and steering torque revealed that part of the muscles produced opposite torque to the steering torque,i.e.,the muscle co-contraction.On this basis,the surface electromyography(EMG)acquisition technology was used to obtain the EMG signal of the mainly activating muscles when the driver performing driving maneuver on the proving ground.The original signals was denoised using wavelet transform.The EMG time domain characteristics were extracted and individual calibration was performed.The results showed that muscle activity and co-contraction were related to the vehicle moving state and the subjective feeling of the driver.Muscle activity was related to the effort which could effectively characterize the driver's physical workload while muscle co-contraction was related to the maneuverability and could effectively characterize the driver's ability to operate.Therefore,muscle co-contraction was used to quantitatively describe the maneuverability of driver's steering maneuver.An evaluation function was established by using the muscle activity of the lower limbs to quantitatively describe the maneuverability of the driver's braking maneuver.At the same time,the natural driving characteristics boundary was established by using the vehicle moving state and the traffic environment characteristics during manual driving,which quantitatively describing the driver's dynamic maneuverability.Thirdly,in order to predict the vehicle moving state during the control authority transitions,the characteristics of vehicle actuators were analyzed and the collision avoidance acceleration and vehicle speed reduction calculation methods were proposed during the braking take-over maneuver.The model predictive control algorithm was used to establish the driver steering control model combined with the musculoskeletal control model based on neuromuscular system characteristics as well as situation awareness.According to the characteristics of the vehicle moving state,the evaluation criteria of control authority transition was established from the aspect of comfort and feasibility.Finally,based on the invariant set theory and the real-time identification method of road traffic environment,the driver's natural driving controllability sets in different traffic environments were established based on natural driving characteristics.The state of the vehicle and the traffic environment during control authority transition was predicted based on reachability analysis to determine whether it met the natural driving characteristics boundary.Considering the evaluation criteria proposed above,a cost function was established to evaluate the overall control authority transition.The cost and of each criterion was calculated based on ANP and entropy weight method after subjective-objective consistency check.The verification was carried out in both simulation and real vehicle environment.The results showed that the method comprehensively considered the influence of the driver-vehicle-road and realized effective quantification of the control authority transition.