Experimental results on combined automatic lane keeping and driver's steering

Automated vehicle steering control systems have great potential to improve road safety. The development of such systems calls for mathematical driver models able to represent human drivers' steering behavior in response to automated steering intervention. This article concerns the experimental evaluation of a game-theoretic driver steering control model. The driver model centers on a steering control strategy developed based on the Nash equilibrium of a theoretic noncooperative game between the driver and automated steering controller. The key parameters of the game-theoretic driver model are identified by fitting the model to real driver steering behavior measured from six driver subjects in an experiment using a driving simulator. The game-theoretic driver model is evaluated by compared to a ``conventional'' optimal-control-theoretic driver model, and analyzing their model fitting errors. Results from the analysis demonstrate that the game-theoretic driver model is statistically significantly better than the conventional driver model for representing three out of the six subjects' steering behavior. For the other three subjects, both the two models perform statistically equivalently well.

It is well established that not only vision but also other sensory modalities affect drivers’ control of their vehicles, and that drivers adapt over time to persistent changes in sensory cues (for example in driving simulators), but the mechanisms underlying these behavioural phenomena are poorly understood. Here, we consider the existing literature on how driver steering in slalom tasks is affected by down-scaling of vestibular cues, and propose, for the first time, a computational model of driver behaviour that can, based on neurobiologically plausible mechanisms, explain the empirically observed effects, namely: decreased task performance and increased steering effort during initial exposure, followed by a partial reversal of these effects as task exposure is prolonged. Unexpectedly, the model also reproduced another previously unexplained empirical finding: a local optimum for motion down-scaling, where path-tracking is better than when one-to-one motion cues are available. Overall, our findings suggest that: (1) drivers make direct use of vestibular information as part of determining appropriate steering actions, and (2) motion down-scaling causes a yaw rate underestimation phenomenon, where drivers behave as if the simulated vehicle is rotating more slowly than it is. However, (3) in the slalom task, a certain degree of such underestimation brings a path-tracking performance benefit. Furthermore, (4) behavioural adaptation in simulated slalom driving tasks may occur due to (a) down-weighting of vestibular cues, and/or (b) increased sensitivity in timing and magnitude of steering corrections, but (c) seemingly not in the form of a full compensatory rescaling of the received vestibular input. The analyses presented here provide new insights and hypotheses about simulated driving and simulator design, and the developed models can be used to support research on multisensory integration and behavioural adaptation in both driving and other task domains.

Automated driving technology is being developed to reduce driver workloads and improve driving safety. However, how connected drivers feel to the control of their vehicles can be a critical factor in driving comfort. In this paper, we discuss a highly automated and human-machine cooperative steering system. Our prototype system regulates driver's steering maneuvers under supervision by automation and also applies guidance torque for physical interaction between a driver and the automation. To evaluate driving comfort and driver behaviors during the cooperation, experiments were conducted with a driving simulator. The results from 20 participants show that drivers who assertively participated in cooperative steering tended to feel safer and more pleasure than in fully automated-like steering.

<div class="htmlview paragraph">Information on the driving stability can be obtained in the form of oversteer or understeer tendencies and acceleration limits on a circular path. Up to now the calculations have only been made on an ideal, even roadway. Significant dynamic characteristics, such as the tuning of the shock absorbers, which also play a role in evaluating the stability of a vehicle in a driving test, are thereby not included in the calculations.</div> <div class="htmlview paragraph">In this paper the unevenness of the roadway is taken into consideration in calculating steady state cornering tests. Different vehicle models can be evaluated in regard to their handling or active safety with the aid of the driver's steering actions required to keep the car on course.</div>

Research and development are actively conducted for recent advances in the self-driving and unmanned vehicles, along with further efficiency and safety. In such fields, obstacle avoidance is frequently addressed in control applications through the concept of active safety technology. This paper investigates the real-time optimal obstacle avoidance control that the optimal path generator embedded model predictive controller. By offline computation, a large number of optimal obstacle avoidance control solutions are calculated to generate the supervised learning data for the deep-learning neural network used for the optimal path generator. Then, constructed optimal path generator is embedded in the model predictive controller to generate the reference command for the tracking task during the optimization process. The utility of the proposed controller is evaluated by both the numerical simulation and experimental testing of the obstacle avoidance maneuver of the micro-unmanned vehicle. Results show that the proposed real-time optimal controller successfully performs obstacle avoidance.

A novel closed-loop control strategy is designed and researched for a matrix converter with dual voltage synthesis control. The closed-loop control strategy is based on the dual-voltage control of the matrix converter duty cycle calculation characteristics, the deviation between the ideal input voltage duty cycle and the equivalent input voltage duty cycle is calculated, and the calculated deviation is added as the negative feedback variable of the closed loop system to the next duty cycle calculation period. In order to achieve the purpose of closed-loop control, the closed-loop control strategy proposed in this paper is used to solve the problem that the output side voltage performance of the matrix converter is affected by the input side voltage distortion and the performance of the internal components of the matrix converter is not ideal. To ensure that the actual output voltage of the matrix converter can better meet the desired voltage, improve its disturbance after the output voltage quality reduction and improve the output performance. The experimental results show that the output voltage of the matrix converter with this new closed - loop control method is closer to the desired voltage, the output current waveform is smoother and the output voltage quality is more ideal.

Modeling driver steering behavior in restricted-preview boundary-avoidance tasks

A simulation investigation of emergency evasive maneuvers of the type experienced during obstacle avoidance is conducted for long combination vehicles at highway speeds. The dynamic model includes a commercial truck with a configuration that is typical of what is used in the U.S., with two 28-ft trailers connected to the tractor through a fifth wheel, and to each other by an A-dolly. The model is validated against test data and used for evaluating distance and time needed to safely perform a steering maneuver to avoid an obstacle on the road. The simulation results indicate a significant difference in both the required time and distance for safely exercising an emergency obstacle avoidance, ranging from approximately 151–276 ft in distance and 1.7–2.4 s from the start of the driver action. The reaction time of the driver will add to both measures by an amount corresponding to the lapse in time between the moment the driver recognises the obstacle and when a steering action is initiated. To the extent that the lighting and road condition add to the reaction time, a significantly larger distance and longer time are needed during nighttime driving as compared with daytime, under similar road and loading conditions.

The aim of this work is to develop a comprehensive yet practical driver model to be used in studying driver–vehicle interactions. Drivers interact with their vehicle and the road through the steering wheel. This interaction forms a closed-loop coupled human–machine system, which influences the driver's steering feel and control performance. A hierarchical approach is proposed here to capture the complexity of the driver's neuromuscular dynamics and the central nervous system in the coordination of the driver's upper extremity activities, especially in the presence of external disturbance. The proposed motor control framework has three layers: the first (or the path planning) plans a desired vehicle trajectory and the required steering angles to perform the desired trajectory; the second (or the musculoskeletal controller) actuates the musculoskeletal arm to rotate the steering wheel accordingly; and the final layer ensures the precision control and disturbance rejection of the motor control units. The physics-based driver model presented here can also provide insights into vehicle control in relaxed and tensed driving conditions, which are simulated by adjusting the driver model parameters such as cognition delay and muscle co-contraction dynamics.

This paper proposes a novel method of verifying comprehensive driver model used for the evaluation of driving safety systems, which is achieved by coupling the traffic simulation and the driving simulator (DS). The method consists of three-step procedure. In the first step, an actual driver operates a DS vehicle in the traffic flow controlled by the traffic simulation. Then in the next step, the actual driver is replaced by a driver model and the surrounding vehicle maneuvers are replayed using the recorded data from the first step. Then, the maneuver by the driver model is compared directly with the actual driver's maneuver along the simulation time steps. For example, while turning right (in front of the oncoming traffic), the proposed verification method was applied to our driver-agent model of traffic simulation system ASSTREET (The Advanced Safety System & Traffic REaltime Evaluation Tool) to demonstrate that the proposed method provides a straightforward way of verifying the driver-agent model capability to reproduce the driving behavior including the subtle driving behaviors, such as safety confirmation, that can have significant impact on driving safety. The scheme of interactive connection of ASSTREET and a driving simulator is also described. Language: en

Integrated Driver and Active Steering Control for Vision-Based Lane Keeping

Swarming UAV flight offers several advantages over single UAV flight. However, it requires a particular guidance paradigm since the collective behavior emerges from the behavior of each individual. This paper proposes a swarming guidance algorithm developed from the artificial potential field concept. The guidance algorithm is capable of aggregation, swarm collision avoidance, waypoint tracking, obstacle avoidance, and rotational maneuvers. The method is simulated using numerical simulation with a quadcopter UAV model.

Autonomous vehicles which are capable of operating independently will be commercially available in the near future. Autonomous driving systems are becoming more complicated and must be successfully checked before implementation. Within this framework, falls our research work. The key purpose of this paper is to implement a simulation environment for autonomous vehicles. We first created this environment which is a novel high fidelity driving simulator that can connect arbitrary interfaces, build simulated worlds consisting of scenarios and incidents experienced by drivers in real-world driving, and incorporate fully autonomous driving. The simulator makes possible to clone the behavior of human driver face as well as some complex situations such as obstacle avoidance maneuvers. The work consists in creating a virtual simulation environment to collect training data used to train vehicles on how to steer themselves. The simulator is, thus, like a video game of car racing. Indeed we used the scenes to make some driving experiences. After collecting the training data, we chose to use deep learning explicitly Convolutional Neural Networks to create a model for autonomous vehicles that avoid obstacles. Clearly, the true challenge for an autonomous vehicle is to navigate without the possibility of collision. This simulator is invested to assess the performance of an autonomous vehicle and to analyze its self-driving activities. In this method, the suggested solution proves to be feasible, efficient and reliable for autonomous vehicle simulation research.

It is well understood that driver's steering input strongly affects lateral vehicle dynamics and excessive steering command may result in unstable vehicle motion. In a certain driving condition, it is possible for a skilled driver to prevent vehicle rollover with better perceptive capability of judging conditions and responding faster with smooth compensatory actions. This paper investigates the possibility of using active steering and wheel torque control to assist drivers in avoiding vehicle rollovers in emergency situations. The effectiveness of steering control alone and combination of steering/wheel torque control in recovery from unstable vehicle roll condition was demonstrated through simulation of both low and high vehicle speeds.

It is well understood that the driver's steering input strongly affects lateral vehicle dynamics, and excessive steering command may result in unstable vehicle motion. In a certain driving condition, it is possible for a skilled driver to prevent vehicle rollover with better perceptive capability of judging conditions and responding faster with smooth compensatory actions. This paper investigates the possibility of using active steering and wheel torque control to assist drivers in avoiding vehicle rollovers in emergency situations. The effectiveness of steering control alone and the combination of steering/wheel torque control in recovery from an unstable roll condition were demonstrated through simulation of both low and high vehicle speeds.