While driving behavior is generally governed by the nature and the driving objectives of the driver, there are many situations (typically in crowded traffic conditions) where tacit communication between vehicle drivers and pedestrians govern driving behavior, significantly influencing transportation safety. The study aimed to formalize the tacit communication between vehicle drivers and pedestrians, in order to inform an investigation on effective communication mechanisms between autonomous vehicle and humans. Current autonomous vehicles engage in decision making primarily controlled by on-board or external sensory information, and do not explicitly consider communication with pedestrians. The study was a within subject 2x2x2 factorial experimental design. The three independent variables were driving context (normal driving vs. autonomous vehicle placard), driving route (1 vs. 2), and narration (yes vs. no). The primary outcome variable was driver-yield behavior. Each of the ten drivers completed the factorial design, requiring eight total drives. Data were collected using a data acquisition system (DAS) designed and installed on the experimental vehicle by the Virginia Tech Transportation Institute. The DAS collected video, audio, and kinematic data. Videos were coded using a proprietary software program, Hawkeye, based on an a priori data directory. Recommendations for future autonomous vehicle research and programming are provided. / Ph. D. / To improve traffic safety and efficiency, the current study examined factors of pedestrian-vehicle interactions. Driving is a dangerous endeavor for all parties, however, pedestrians are an especially vulnerable group. Many different solutions have been suggested including; education and training of road users, high visibility law enforcement, infrastructure changes, and vehicle solutions. Of all proposed, the vehicle solution, autonomous vehicles, shows great promise in improving traffic safety. Autonomous vehicles provide an opportunity for a high degree of safety, yet, inefficiencies exist. For instance, a vehicle might stop at all crosswalks regardless of pedestrian proximity. To this end, the current study was a scientific exploration of the factors relating to pedestrian-vehicle interactions. The exploratory nature of this work provided an opportunity to provide recommendations for programming of autonomous vehicles to balance safety and efficiency.

Vehicle fuel consumption and emission are two important effectiveness measurements of sustainable transportation development. Pavement plays an essential role in goals of fuel economy improvement and greenhouse gas (GHG) emission reduction. The main objective of this dissertation study is to experimentally investigate the effect of pavement-vehicle interaction (PVI) on vehicle fuel consumption under highway driving conditions. The goal is to provide a better understanding on the role of pavement in the green transportation initiates. Four study phases are carried out. The first phase involves a preliminary field investigation to detect the fuel consumption differences between paired flexible-rigid pavement sections with repeat measurements. The second phase continues the field investigation by a more detailed and comprehensive experimental design and independently investigates the effect of pavement type on vehicle fuel consumption. The third study phase calibrates the HDM-IV fuel consumption model with data collected in the second field phase. The purpose is to understand how pavement deflection affects vehicle fuel consumption from a mechanistic approach. The last phase applies the calibrated HDM-IV model to Florida’s interstate network and estimates the total annual fuel consumption and CO2 emissions on different scenarios. The potential annual fuel savings and emission reductions are derived based on the estimation results. Statistical results from the two field studies both show fuel savings on rigid pavement compared to flexible pavement with the test conditions specified. The savings derived from the first phase are 2.50% for the passenger car at 112km/h, and 4.04% for 18-wheel tractor-trailer at 93km/h. The savings resulted from the second phase are 2.25% and 2.22% for passenger car at 93km/h and 112km/h, and 3.57% and 3.15% for the 6-wheel medium-duty truck at 89km/h and 105km/h. All savings are statistically significant at 95% Confidence Level (C.L.). From the calibrated HDM-IV model, one unit of pavement deflection (1mm) on flexible pavement can cause an excess fuel consumption by 0.234-0.311 L/100km for the passenger car and by 1.123-1.277 L/100km for the truck. The effect is more evident at lower highway speed than at higher highway speed. From the network level estimation, approximately 40 million gallons of fuel (combined gasoline and diesel) and 0.39 million tons of CO2 emission can be saved/reduced annually if all Florida’s interstate flexible pavement are converted to rigid pavement with the same roughness levels. Moreover, each 1-mile of flexible-rigid conversion can result in a reduction of 29 thousand gallons of fuel and 258 tons of CO2 emission yearly.

Fully Autonomous Vehicles (FAVs) have the potential to provide safer vehicle operation and to enhance the overall transportation system. However, drivers and vehicles are not the only components that need to be considered. Research has shown that pedestrians are among the most unpredictable and vulnerable road users. To achieve full and successful implementation of FAVs, it is essential to understand pedestrian acceptance and intended behavior regarding FAVs. Three studies were developed to address this need: (1) development of a standardized framework to investigate pedestrians’ behaviors for the U.S. population; (2) development of a framework to evaluate their receptivity of FAVs; and (3) investigation of the influence of the external interacting interfaces of FAVs on pedestrian receptivity toward them. The pedestrian behavior questionnaire (PBQ) categorized pedestrian general behaviors into five factors: violations, errors, lapses, aggressive behaviors, and positive behaviors. The first four factors were found to be both valid and reliable; the positive behavior scale was not found to be reliable nor valid. A long (36-item) and a short (20-items) versions of the PBQ were validated by regressing scenario-based survey responses to the fiveactor PBQ subscale scores. The pedestrian receptivity questionnaire for FAVs (PRQF) consisted of three subscales: safety, interaction, and compatibility. This factor structure was verified by a confirmatory factor analysis and the reliability of each subscale was confirmed. Regression analyses showed that pedestrians’ intention to cross the road in front of a FAV was significantly predicted by both safety and interaction scores, but not by the compatibility score. On the other hand, acceptance of FAVs in the existing traffic system was predicted by all three subscale scores. Finally, an experimental study was performed to expose pedestrians to a simulated environment where they could experience a FAV. The FAV in the simulated environment was either equipped with external features (audible and/or visual) or had no external (warning) feature. The least preferred options were the FAVs with no features and those with a smiley face but no audible cue. The most preferred interface option, which instilled confidence for crossing in front of the FAV, was the walking silhouette.

No description available.

Human activity recognition is one of the intensively studied areas in computer vision. Most existing works do not assume video resolution to be a problem due to general applications of interests. However, with continuous concerns about global security and emerging needs for intelligent video analysis tools, activity recognition from low-resolution and low-quality videos has become a crucial topic for further research. In this dissertation, We present a series of approaches which are developed specifically to address the related issues regarding low-level image preprocessing, single person activity recognition, and human-vehicle interaction reasoning from low-resolution surveillance videos. Human cast shadows are one of the major issues which adversely effect the performance of an activity recognition system. This is because human shadow direction varies depending on the time of the day and the date of the year. To better resolve this problem, we propose a shadow removal technique which effectively eliminates a human shadow cast from a light source of unknown direction. A multi-cue shadow descriptor is employed to characterize the distinctive properties of shadows. Our approach detects, segments, and then removes shadows. We propose two different methods to recognize single person actions and activities from low-resolution surveillance videos. The first approach adopts a joint feature histogram based representation, which is the concatenation of subspace projected gradient and optical flow features in time. However, in this problem, the use of low-resolution, coarse, pixel-level features alone limits the recognition accuracy. Therefore, in the second work, we contributed a novel mid-level descriptor, which converts an activity sequence into simultaneous temporal signals at body parts. With our representation, activities are recognized through both the local video content and the short-time spectral properties of body parts' movements. We draw the analogies between activity and speech recognition and show that our speech-like representation and recognition scheme improves recognition performance in several low-resolution datasets. To complete the research on this subject, we also tackle the challenging problem of recognizing human-vehicle interactions from low-resolution aerial videos. We present a temporal logic based approach which does not require training from event examples. At the low-level, we employ dynamic programming to perform fast model fitting between the tracked vehicle and the rendered 3-D vehicle models. At the semantic-level, given the localized event region of interest (ROI), we verify the time series of human-vehicle spatial relationships with the pre-specified event definitions in a piecewise fashion. Our framework can be generalized to recognize any type of human-vehicle interaction from aerial videos. / text

With the growing complexity of vehicle control systems it is becoming increasingly important to understand the interaction between drivers and vehicles. Existing driver models do not adequately characterise limitations resulting from drivers’ physical systems. In particular, sensory dynamics limit the ability of drivers to perceive the states of real or simulated vehicles. Therefore, the aim of this thesis is to understand the impact of sensory dynamics on the control performance of a human driver in real and virtual environments. A new model of driver steering control is developed based on optimal control and state estimation theory, incorporating models of sensory dynamics, delays and noise. Some results are taken from published literature, however recent studies have shown that sensory delays and noise amplitudes may increase during an active control task such as driving. Therefore, a parameter identification procedure is used to fit the model predictions to measured steering responses of real drivers in a simulator. The model is found to fit measured results well under a variety of conditions. An initial experiment is designed with the physical motion of the simulator matching the motion of the virtual vehicle at full scale. However, during more realistic manoeuvres the physical motion must be scaled or filtered, introducing conflicts between measurements from different sensory systems. Drivers are found to adapt to simple conflicts such as scaled motion, but they have difficulty adapting to more complicated motion filters. The driver model is initially derived for linear vehicles with stochastic target and disturbance signals. In later chapters it is extended to account for transient targets and disturbances and vehicles with nonlinear tyres, and validated once again with experimental results. A series of simulations is used to demonstrate novel insights into how drivers use sensory information, and the resulting impact on control performance. The new model is also shown to predict difficulties real drivers have controlling unstable vehicles more reliably than existing driver models.

abstract: Despite the various driver assistance systems and electronics, the threat to life of driver, passengers and other people on the road still persists. With the growth in technology, the use of in-vehicle devices with a plethora of buttons and features is increasing resulting in increased distraction. Recently, speech recognition has emerged as an alternative to distraction and has the potential to be beneficial. However, considering the fact that automotive environment is dynamic and noisy in nature, distraction may not arise from the manual interaction, but due to the cognitive load. Hence, speech recognition certainly cannot be a reliable mode of communication. The thesis is focused on proposing a simultaneous multimodal approach for designing interface between driver and vehicle with a goal to enable the driver to be more attentive to the driving tasks and spend less time fiddling with distractive tasks. By analyzing the human-human multimodal interaction techniques, new modes have been identified and experimented, especially suitable for the automotive context. The identified modes are touch, speech, graphics, voice-tip and text-tip. The multiple modes are intended to work collectively to make the interaction more intuitive and natural. In order to obtain a minimalist user-centered design for the center stack, various design principles such as 80/20 rule, contour bias, affordance, flexibility-usability trade-off etc. have been implemented on the prototypes. The prototype was developed using the Dragon software development kit on android platform for speech recognition. In the present study, the driver behavior was investigated in an experiment conducted on the DriveSafety driving simulator DS-600s. Twelve volunteers drove the simulator under two conditions: (1) accessing the center stack applications using touch only and (2) accessing the applications using speech with offered text-tip. The duration for which user looked away from the road (eyes-off-road) was measured manually for each scenario. Comparison of results proved that eyes-off-road time is less for the second scenario. The minimalist design with 8-10 icons per screen proved to be effective as all the readings were within the driver distraction recommendations (eyes-off-road time < 2sec per screen) defined by NHTSA. / Dissertation/Thesis / Masters Thesis Computer Science 2015

No description available.

The aim of electrification of chassis and driveline systems in road vehicles is to reduce the global emissions and their impact on the environment. The electrification of such systems in vehicles is enabling a whole new set of functionalities improving safety, handling and comfort for the user. This trend is leading to an increased number of elements in road vehicles such as additional sensors, actuators and software codes. As a result, the complexity of vehicle components and subsystems is rising and has to be handled during operation. Hence, the probability of potential faults that can lead to component or subsystem failures deteriorating the dynamic behaviour of road vehicles is becoming higher. Mechanical, electric, electronic or software faults can cause these failures independently or by mutually influencing each other, thereby leading to potentially critical traffic situations or even accidents. There is a need to analyse faults regarding their influence on the dynamic behaviour of road vehicles and to investigate their effect on the driver-vehicle interaction and to find new control strategies for fault handling. A structured method for the classification of faults regarding their influence on the longitudinal, lateral and yaw motion of a road vehicle is proposed. To evaluate this method, a broad failure mode and effect analysis was performed to identify and model relevant faults that have an effect on the vehicle dynamic behaviour. This fault classification method identifies the level of controllability, i.e. how easy or difficult it is for the driver and the vehicle control system to correct the disturbance on the vehicle behaviour caused by the fault. Fault-tolerant control strategies are suggested which can handle faults with a critical controllability level in order to maintain the directional stability of the vehicle. Based on the principle of control allocation, three fault-tolerant control strategies are proposed and have been evaluated in an electric vehicle with typical faults. It is shown that the control allocation strategies give a less critical trajectory deviation compared to an uncontrolled vehicle and a regular electronic stability control algorithm. An experimental validation confirmed the potential of this type of fault handling using one of the proposed control allocation strategies. Driver-vehicle interaction has been experimentally analysed during various failure conditions with typical faults of an electric driveline both at urban and motorway speeds. The driver reactions to the failure conditions were analysed and the extent to which the drivers could handle a fault were investigated. The drivers as such proved to be capable controllers by compensating for the occurring failures in time when they were prepared for the eventuality of a failure. Based on the experimental data, a failure-sensitive driver model has been developed and evaluated for different failure conditions. The suggested fault classification method was further verified with the conducted experimental studies. The interaction between drivers and a fault-tolerant control system with the occurrence of a fault that affects the vehicle dynamic stability was investigated further. The control allocation strategy has a positive influence on maintaining the intended path and the vehicle stability, and supports the driver by reducing the necessary corrective steering effort. This fault-tolerant control strategy has shown promising results and its potential for improving traffic safety. / <p>QC 20150520</p>

This dissertation describes two distinctive methods for human-vehicle interaction recognition: one for ground level videos and the other for aerial videos. For ground level videos, this dissertation presents a novel methodology which is able to estimate a detailed status of a scene involving multiple humans and vehicles. The system tracks their configuration even when they are performing complex interactions with severe occlusion such as when four persons are exiting a car together. The motivation is to identify the 3-D states of vehicles (e.g. status of doors), their relations with persons, which is necessary to analyze complex human-vehicle interactions (e.g. breaking into or stealing a vehicle), and the motion of humans and car doors to detect atomic human-vehicle interactions. A probabilistic algorithm has been designed to track humans and analyze their dynamic relationships with vehicles using a dynamic context. We have focused on two ideas. One is that many simple events can be detected based on a low-level analysis, and these detected events must contextually meet with human/vehicle status tracking results. The other is that the motion clue interferes with states in the current and future frames, and analyzing the motion is critical to detect such simple events. Our approach updates the probability of a person (or a vehicle) having a particular state based on these basic observed events. The probabilistic inference is made for the tracking process to match event-based evidence and motion-based evidence. For aerial videos, the object resolution is low, the visual cues are vague, and the detection and tracking of objects is less reliable as a consequence. Any method that requires accurate tracking of objects or the exact matching of event definition are better avoided. To address these issues, we present a temporal logic based approach which does not require training from event examples. At the low-level, we employ dynamic programming to perform fast model fitting between the tracked vehicle and the rendered 3-D vehicle models. At the semantic-level, given the localized event region of interest (ROI), we verify the time series of human-vehicle relationships with the pre-specified event definitions in a piecewise fashion. With special interest in recognizing a person getting into and out of a vehicle, we have tested our method on a subset of the VIRAT Aerial Video dataset and achieved superior results. / text