Search Results

This paper explores the role and challenges of Artificial Intelligence (AI) algorithms, specifically AI-based software elements, in autonomous driving systems. These AI systems are fundamental in executing real-time critical functions in complex and high-dimensional environments. They handle vital tasks like multi-modal perception, cognition, and decision-making tasks such as motion planning, lane keeping, and emergency braking. A primary concern relates to the ability (and necessity) of AI models to generalize beyond their initial training data. This generalization issue becomes evident in real-time scenarios, where models frequently encounter inputs not represented in their training or validation data. In such cases, AI systems must still function effectively despite facing distributional or domain shifts. This paper investigates the risk associated with overconfident AI models in safety-critical applications like autonomous driving.

The advancement of Machine-learning (ML) methods enables data-driven creation of Reduced Order Models (ROMs) for automotive components and systems. For example, Gaussian Process Regression (GPR) has emerged as a powerful tool in recent years for building a static ROM as an alternative to a conventional parametric model or a multi-dimensional look-up table. GPR provides a mathematical framework for probabilistically representing complex non-linear behavior. Today, GPR is available in various programing tools and commercial CAE packages. However, the application of GPR is system dependent and often requires careful design considerations such as selection of input features and specification of kernel functions. Hence there is a need for GPR design optimization driven by application requirements. For example, a moving window size for training must be tuned to balance performance and computational efficiency for tracking changing system behavior.

A wet clutch is an established component in a conventional powertrain. It also finds a new role in electrified systems. For example, a wet clutch is utilized to couple or decouple an internal combustion engine from an electrically-driven drivetrain on demand in hybrid electric vehicles. In some electrical vehicle designs, it provides a means for motor speed reduction. Wet clutch control for those new applications may differ significantly from conventional strategy. For example, actuator pressure may be heavily modulated, causing the clutch to exhibit pronounced hysteresis. The clutch may be required to operate at a very high slip speed for unforeseen behaviors. A linear transfer function is commonly utilized for clutch control in automating shifting applications, assuming that clutch torque is proportional to actuator pressure. However, the linear model becomes inadequate for enabling robust control when the clutch behavior becomes highly nonlinear with hysteresis.

Electrically assisted turbochargers are a promising technology for improving boost response of turbocharged engines. These systems include a turbocharger shaft mounted electric motor/generator. In the assist mode, electrical energy is applied to the turbocharger shaft via the motor function, while in the regenerative mode energy can be extracted from the shaft via the generator function, hence these systems are also referred to as regenerative electrically assisted turbochargers (REAT). REAT allows simultaneous improvement of boost response and fuel economy of boosted engines. This is achieved by optimally scheduling the electrical assist and regeneration actions. REAT also allows the exhaust turbine to operate within a narrow range of optimal vane positions relative to the unassisted variable geometry turbocharger (VGT). The ability to operate within a narrow range of VGT vane positions allows an opportunity for a more optimal turbine design for a REAT system.

The availability of connectivity and autonomy enabled resources, within the automotive sector, has primarily been considered for driver assist technologies and for extending the levels of vehicle autonomy. It is not a stretch to imagine that the additional information, available from connectivity and autonomy, may also be useful in further improving powertrain functions. Critical powertrain subsystems that must operate with limited or uncertain knowledge of their environment stand to benefit from such new information sources. Unfortunately, the adoption of this new information resource has been slow within the powertrain community and has typically been limited to the obvious problem choices such as battery charge management for electric vehicles and efforts related to fuel economy benefits from adaptive/coordinated cruise control. In this paper we discuss the application of connectivity resources in the management of an aftertreatment sub-system, the Diesel Particulate Filter (DPF).

Over the past decade urea-based selective catalytic reduction (SCR) has become a leading aftertreatment solution to meet increasingly stringent Nitrogen oxide (NOx) emissions requirements in diesel powertrains. A common trend seen in modern SCR systems is the use of "split-brick" configurations where two SCR catalysts are placed in thermally distinct regions of the aftertreatment. One catalyst is close-coupled to the engine for fast light-off and another catalyst is positioned under-floor to improve performance at high space velocities. Typically, a single injector is located upstream of the first catalyst to provide the reductant necessary for efficient NOx reduction. This paper explores the potential benefit, in terms of improved NOx reduction, control of NH3 slip or reduced reductant consumption, of having independently actuated injectors in front of each catalyst.

Many excellent papers have been written about the subject of estimating engine-out NOx on diesel engines based on real-time available data. The claimed accuracy of these models is typically around 6-10% on validation data sets with known inputs. This reported accuracy typically ignores input uncertainties, thus arriving at an optimistic estimate of the model accuracy in a real-time application. In our paper we analyze the effect of input uncertainty on the accuracy of engine-out NOx estimates via a numerical Monte Carlo simulation and show that this effect can be significant. Even though our model is based on an in-cylinder pressure sensor, this sensor is limited in its capability to reduce the effect of other measured inputs on the model.

Ammonia adsorption on the catalyst surface is a crucial step in the selective catalytic reduction of nitrogen oxides over zeolites with NH3 as the reducing agent. In this study, two small pore zeolites with chabazite frameworks, H-SSZ-13 and Cu exchanged SSZ-13, are examined. Adsorption of NH3 on the zeolite causes changing electrical properties of the material. They can be detected by a radio frequency based technique. We have found that with this method it is possible to determine the amount of adsorbed NH3 on these catalysts, examining both the influences of temperature and NH3/NO feed gas ratio. At constant temperature, a fairly linear correlation between the resonance frequency and the amount of adsorbed ammonia was observed. Furthermore, this method also allows differentiation between some of the NH3 adsorption sites.

Urea selective catalytic reduction (SCR) catalysts have the capability to deliver the high NOx conversion efficiencies required for future emission standards. However, the potential for the occasional over-temperature can lead to the irreversible deactivation of the SCR catalyst. On-board diagnostics (OBD) compliance requires monitoring of the SCR function to make sure it is operating properly. Initially, SCR catalyst performance metrics such as NOx conversion, NH3 oxidation, NH3 storage capacity, and BET surface area are within normal limits. However, these features degrade with high temperature aging. In this work, a laboratory flow reactor was utilized to determine the impact on these performance metrics as a function of aging condition. Upon the completion of a full time-at-temperature durability study, four performance criteria were established to help determine a likely SCR failure.

Regulatory authorities are actively revising and updating the rules for on board diagnostics of diesel powertrains. Diesel oxidation catalysts are among the parts that will have to be monitored. This paper discusses some of the issues related to the feasibility of monitoring these catalysts. We concentrate on the effect of real world noise factors on the ability to distinguish marginal from threshold catalysts and demonstrate that with current sensor and catalyst technology the separation between the two is poor.

Diesel particulate filters (DPFs) are a technology likely to be deployed to meet future stringent emission levels for heavy and light duty diesel powertrains in North America and Europe. This paper discusses experimental results in the active regeneration of DPFs. Attention is given to the system components, the information based on which regeneration is triggered, and the means to achieve a regeneration. The paper will report on successful regenerations under several extreme conditions.

Carbon canisters have been adapted for automobile use since the early 1970s to control evaporative emissions. Stringent emission regulations and the requirement for an enhanced evaporative emissions test procedure, make this an important issue. The air and evaporative fuel from the carbon canister therefore need important consideration with respect to air to fuel ratio (AFR) control and idle by-pass air control. Although a few complex models of the activated carbon canister have been developed, a control-oriented, simplistic carbon canister model needs to be developed. This paper explores the control-oriented modeling of a canister purge air system along with the on-line estimation of evaporative fuel loading of the activated carbon. An attempt was made at providing an analytical expression for the evaporative fuel and air entering the intake manifold.

Due to the stringent emission regulations, On-Board Diagnostics II (OBD II) and the requirement of enhanced evaporative emissions test procedure, an aggressive canister purge control strategy is required for automotive vehicles. The enhanced evaporative emissions test procedure has forced car manufacturer to purge the carbon canister in the vehicle idle condition so that production vehicles meet the SHED and hot soak test requirements. This not only worsens the idle speed quality but also tends to increase exhaust emission levels. Using analytical models of evaporative air and fuel, feed-forward control strategy for both idle by-pass air and air to fuel ratio can be improved. This paper demonstrates an application of evaporative system modeling to the idle air and air to fuel ratio control.

Over the years numerous researchers have suggested that the ionization current signal carries within it combustion relevant information. The possibility of using this signal for diagnostics and control provides motivation for continued research in this area. To be able to use the ion current signal for feedback control a reliable estimate of some combustion related parameter is necessary and therein lies the difficulty. Given the nature of the ion current signal this is not a trivial task. Fei An et al. [1] employed PCA for feature extraction and then used these feature vectors to design a neural network based classifier for the estimation of air to fuel ratio (AFR). Although the classifier predicted AFR with sufficient reliability, a major draw back was that the ion current signals used for prediction were averaged signals thus precluding a cycle to cycle estimate of AFR.

The process of incorporating the spark plug as a combustion probe, to perform misfire and knock detection, air to fuel ratio and spark timing control has been the subject of research for some time now. [3], [4]. The feasibility of the approach however depends on being able to correlate some characteristic of the ion current signal to the in cylinder combustion process. Shimaski et al. [3] and Miyata et al. [4] suggest such a relationship. The objective of this research has been to extract combustion information from the measured ion current flowing between spark plug electrodes by using various advanced signal processing methods, and to develop a methodology that will permit combustion diagnostics and possibly control based on these measurements. Tests were carried out on a single-cylinder, methane-fueled CFR engine.