Search Results

This paper seeks to identify key input parameters needed to achieve a production-viable control strategy for spark-assisted compression ignition (SACI) engines. SACI is a combustion strategy that uses a spark plug to initiate a deflagration flame that generates sufficient ignition energy to trigger autoignition in the remaining charge. The flame propagation phase limits the rate of cylinder pressure rise, while autoignition rapidly completes combustion. High dilution within the autoignited charge is generally required to maintain reaction rates feasible for production. However, this high dilution may not be reliably ignited by the spark plug. These competing constraints demand novel mixture preparation strategies for SACI to be feasible in production. SACI with charge stratification has demonstrated sufficiently stable flame propagation to reliably trigger autoignition across much of the engine operating map.

The deployment of autonomous vehicles in real-world scenarios requires thorough testing to ensure sufficient safety levels. Driving simulators have proven to be useful testbeds for assisted and autonomous driving functionalities but may fail to capture all the nuances of real-world conditions. In this paper, we present a snapshot of the design and evaluation using a Cooperative Adaptive Cruise Control application of virtual reality platform currently in development at our institution. The platform is designed so to: allow for incorporating live real-world driving data into the simulation, enabling Vehicle-in-the-Loop testing of autonomous driving behaviors and providing us with a useful mean to evaluate the human factor in the autonomous vehicle context.

In this paper, we present a framework for the robot to learn how to place objects to a workpiece by learning from humans in smart manufacturing. In the proposed framework, the rational scene dictionary (RSD) corresponding to the keyframes of task (KFT) are used to identify the general object-action-location relationships. The Generalized Voronoi Diagrams (GVD) based contour is used to determine the relative position and orientation between the object and the corresponding workpiece at the final state. In the learning phase, we keep tracking the image segments in the human demonstration. For the moment when a spatial relation of some segments are changed in a discontinuous way, the state changes are recorded by the RSD. KFT is abstracted after traversing and searching in RSD, while the relative position and orientation of the object and the corresponding mount are presented by GVD-based contours for the keyframes.

Unmanned ground vehicles (UGVs) may encounter difficulties accommodating environmental uncertainties and system degradations during harsh conditions. However, human experience and onboard intelligence can may help mitigate such cases. Unfortunately, human operators have cognition limits when directly supervising multiple UGVs. Ideally, an automated decision aid can be designed that empowers the human operator to supervise the UGVs. In this paper, we consider a connected UGV platoon under cyber attacks that may disrupt safety and degrade performance. An observer-based resilient control strategy is designed to mitigate the effects of vehicle-to-vehicle (V2V) cyber attacks. In addition, each UGV generates both internal and external evaluations based on the platoons performance metrics. A cloud-based trust-based information management system collects these evaluations to detect abnormal UGV platoon behaviors.

Automotive drive-by-wire systems have enabled greater mobility options for individuals with physical disabilities. To further expand the driving paradigm, a need exists to consider an alternative vehicle steering mechanism to meet specific needs and constraints. In this study, a cellphone steering controller was investigated using a fixed-base driving simulator. The cellphone incorporated the direction control of the vehicle through roll motion, as well as the brake and throttle functionality through pitch motion, a design that can assist disabled drivers by excluding extensive arm and leg movements. Human test subjects evaluated the cellphone with conventional vehicle control strategy through a series of roadway maneuvers. Specifically, two distinctive driving situations were studied: a) obstacle avoidance test, and b) city road traveling test. A conventional steering wheel with self-centering force feedback tuning was used for all the driving events for comparison.

It’s important to predict human actions in the industry assembly process. Foreseeing future actions before they happened is an essential part for flexible human-robot collaboration and crucial to safety issues. Vision-based human action prediction from videos provides intuitive and adequate knowledge for many complex applications. This problem can be interpreted as deducing the next action of people from a short video clip. The history information needs to be considered to learn these relations among time steps for predicting the future steps. However, it is difficult to extract the history information and use it to infer the future situation with traditional methods. In this scenario, a model is needed to handle the spatial and temporal details stored in the past human motions and construct the future action based on limited accessible human demonstrations.

Non-linear model predictive engine control (nMPC) systems have the ability to reduce calibration effort while improving transient engine response. The main drawback of nMPC for engine control is the computational power required to realize real-time operation. Most of this computational power is spent linearizing the non-linear plant model at each time step. Additionally, the effectiveness of the nMPC system relies heavily on the accuracy of the model(s) used to predict the future system behavior, which can be difficult to model physically. This paper introduces a hybrid modeling approach for internal combustion engines that combines physics-based and machine learning techniques to generate accurate models that can be linearized with low computational power. This approach preserves the generalization and robustness of physics-based models, while maintaining high accuracy of data-driven models. Advantages of applying the proposed model with nMPC are discussed.

Energy management of hybrid vehicle has been a widely researched area. Strategies like dynamic programming (DP), equivalent consumption minimization strategy (ECMS), Pontryagin’s minimum principle (PMP) are well analyzed in literatures. However, the adaptive optimization work is still lacking, especially for reinforcement learning (RL). In this paper, Q-learning, as one of the model-free reinforcement learning method, is implemented in a mid-size 48V mild parallel hybrid electric vehicle (HEV) framework to optimize the fuel economy. Different from other RL work in HEV, this paper only considers vehicle speed and vehicle torque demand as the Q-learning states. SOC is not included for the reduction of state dimension. This paper focuses on showing that the EMS with non-SOC state vectors are capable of controlling the vehicle and outputting satisfactory results. Electric motor torque demand is chosen as action.

This paper describes the development and simulation of a voice and pointing gesture interaction system for on-route update of autonomous vehicles’ path. The objective of this research is to provide users of autonomous vehicles a human vehicle interaction mode that enables them to make and communicate spontaneous decisions to the autonomous car, modifying its pre-defined autonomous route in real-time. For example, similar to giving directions to a taxi driver, a user will be able to tell the car «Stop there» or «Take that exit». In this way, the user control/spontaneity vs interaction flexibility dilemma that current autonomous vehicle concepts have, could be solved, potentially increasing the user acceptance of this technology. The system was designed following a level structured state machine approach. The simulations were developed using MATLAB and VREP, a robotics simulation platform, which has accurate vehicle and sensor models.

Fuel efficiency improvement in automobiles has been a topic of great interest over the past few years, especially with the introduction of the new CAFE 2025 standards. Although there are multiple ways of improving the fuel efficiency of an automobile, lightweighting is one of the most common approaches taken by many automotive manufacturers. Lightweighting is even more significant in electric vehicles as it directly affects the range of the vehicle. Amidst this context of lightweighting, the use of composite materials as alternatives to metals has been proven in the past to help achieve substantial weight reduction. The focus of using composites for weight reduction has however been typically limited to major structural components, such as BiW and closures, due to high material costs. Secondary structural components which contribute approximately 30% of the vehicle weight are usually neglected by these weight reduction studies.

This study was driven by the prevalence of older drivers’ overrepresentation in crashes caused by pedal application errors. Previous research has shown tasks prone to pedal errors, which include emergency braking, parking lot maneuvers and reaching out of the driver’s window. However, pedal usage characteristics of older drivers while performing on-road driving tasks are unknown. The objective of this research was to understand pedal usage characteristics of older drivers during on-road driving tasks in an instrumented vehicle. Twenty-six drivers over the age of 60 completed 10 stopping tasks as the baseline for stopping performance, a startle-braking task, two forward parking tasks and two reaching out of the vehicle tasks. Results for this instrumented vehicle study showed significantly positive correlations between stature and the percent of foot pivoting, and between shoe length and percent of foot pivoting in the baseline stopping tasks.

Learning only from a small set of examples remains a huge challenge in machine learning. Despite recent breakthroughs in the applications of neural networks, the applicability of these techniques has been limited by the requirement for large amounts of training data. What’s more, the standard supervised machine learning method does not provide a satisfactory solution for learning new concepts from little data. However, the ability to learn enough information from few samples has been demonstrated in humans. This suggests that humans may make use of prior knowledge of a previously learned model when learning new ones on a small amount of training examples. In the area of autonomous driving, the model learns to drive the vehicle with training data from humans, and most machine learning based control algorithms require training on very large datasets. Collecting and constructing training data set takes a huge amount of time and needs specific knowledge to gather relevant information.

The Deep Orange program immerses automotive engineering students into the world of an OEM as part of their 2-year graduate education. In support of developing the program’s seventh vehicle concept, the students studied the sponsoring brand essence, conducted market research, and made a heuristic assessment of competitor vehicles. The upfront research lead to the definition of target customers and setting vehicle level targets that were broken down into requirements to develop various vehicle sub-systems. The powertrain team was challenged to develop a scalable propulsion concept enabled by a common vehicle architecture that allowed future customers to select (at the point of purchase) among various levels of electrification best suiting their needs and personal desires. Four different configurations were identified and developed: all-electric, two plug-in hybrid electric configurations, and an internal combustion engine only.

Improper fit in a vehicle will affect a driver’s ability to reach the steering wheel and pedals, view the roadway and instrument gauges, and allow vehicle safety features to protect the driver during a crash. CarFit® is a community outreach program to educate older drivers on proper “fit” within their personal vehicle. A subset of measurements from CarFit® were used to quantify the “fit” of 97 older drivers over 60 and 20 younger drivers, ages 30-39, in their personal vehicles. Binary, logistic regression was used to assess the likelihood of drivers meeting the CarFit® measurement criteria prior to and after CarFit® education. The results showed older drivers were five times more likely than younger drivers to meet the CarFit® criteria for line of sight above the steering wheel, suggesting that younger drivers would also benefit from CarFit® education.

The ability to independently transfer into and out of a vehicle is essential for many wheelchair users to achieve driving independence. The purpose of the current study is to build upon the previous exploratory study that investigated the transfer strategies of wheelchair users by observing YouTube videos. This observational study videotaped five wheelchair users transferring from their wheelchairs into two research vehicles, a small and mid-size sedan that were equipped with a 50mm grid. The goal of this study was to use these videos and vehicle grids to precisely identify ingress and egress motions as well as “touch points” in a controlled setting with a small sample of five male wheelchair users. Using the videos from multiple different camera perspectives, the participants’ ingress and egress transfers were coded, documenting the touch points and step-by-step action sequences.

This research proposes a control system for Spark Ignition (SI) engines with external Exhaust Gas Recirculation (EGR) based on model predictive control and a disturbance observer. The proposed Economic Nonlinear Model Predictive Controller (E-NMPC) tries to minimize fuel consumption for a number of engine cycles into the future given an Indicated Mean Effective Pressure (IMEP) tracking reference and abnormal combustion constraints like knock and combustion variability. A nonlinear optimization problem is formulated and solved in real time using Sequential Quadratic Programming (SQP) to obtain the desired control actuator set-points. An Extended Kalman Filter (EKF) based observer is applied to estimate engine states, combining both air path and cylinder dynamics. The EKF engine state(s) observer is augmented with disturbance estimation to account for modeling errors and/or sensor/actuator offset.

In the next 20 years fully autonomous vehicles are expected to be in the market. The advance on their development is creating paradigm shifts on different automotive related research areas. Vehicle interiors design and human vehicle interaction are evolving to enable interaction flexibility inside the cars. However, most of today’s vehicle manufacturers’ autonomous car concepts maintain the steering wheel as a control element. While this approach allows the driver to take over the vehicle route if needed, it causes a constraint in the previously mentioned interaction flexibility. Other approaches, such as the one proposed by Google, enable interaction flexibility by removing the steering wheel and accelerator and brake pedals. However, this prevents the users to take control over the vehicle route if needed, not allowing them to make on-route spontaneous decisions, such as stopping at a specific point of interest.

Knock-limited engine operation is one of the most important constraints on fuel efficiency and performance that must be considered during the design, control algorithm development and calibration of spark-ignition engines. This research evaluates the accuracy of model-based knock prediction routines and their applicability for control-oriented applications over various engine operating conditions using commercial fuels. Two common methods of knock prediction, a generalized chemical kinetics model and an empirical induction-time correlation, are evaluated and compared against experimental data. The experimental investigation is conducted using a naturally aspirated 3.6L V6 engine, retrofitted with cooled Exhaust Gas Recirculation (EGR). Data are acquired from spark timing sweeps under knocking conditions at different engine speeds and loads in an engine dynamometer cell.

Autonomous driving technologies can provide better safety, comfort and efficiency for future transportation systems. Most research in this area has mainly been focused on developing sensing and control approaches to achieve various autonomous driving functions. Very little of this research, however, has studied how to efficiently handle sensing exceptions. A simple exception measured by any of the sensors may lead to failures in autonomous driving functions. The autonomous vehicles are then supposed to be sent back to manufacturers for repair, which takes both time and money. This paper introduces an efficient approach to make human drivers able to online teach autonomous vehicles to drive under sensing exceptions. A human-vehicle teaching-and-learning framework for autonomous driving is proposed and the human teaching and vehicle learning processes for handling sensing exceptions in autonomous vehicles are designed in detail.

The duration over which a three way catalyst (TWC) maintains proper functionality during lambda excursions is critically impacted by aging, which affects its oxygen storage capacity (OSC). As such, emissions control strategies, which strive to maintain post TWC air-to-fuel ratios at the stoichiometric value, will benefit from an accurate estimation of TWC age. To this end, this investigation examines a method of TWC age estimation suitable for real-world transient operation. Experimental results are harvested from an instrumented test vehicle equipped with a two-brick TWC during operation on a chassis dynamometer. Four differently aged TWCs are instrumented with wideband and switch-type Lambda sensors upstream (Pre TWC location), and downstream (Mid location) of first catalyst brick.