Search Results

Now-a-days, Advanced driver-assistance systems (ADAS) is equipping cars and drivers with advance information and technology to make them become aware of the environment and handle potential situations in better way semi-autonomously. High-quality training and test data is essential in the development and validation of ADAS systems which lay the foundation for autonomous driving technology. ADAS uses the technology like radar, vision and combinations of various sensors including LIDAR to automatize dynamic driving tasks like steering, braking, and acceleration of vehicle for controlled and safe driving. And to integrate these advance technologies, the ADAS needs labeled data to train the algorithm to detect the various objects and moments of driver. Image annotation is one the well-known service to create such training data for computer vision. There are number of open source annotated datasets available viz. COCO, KITTI etc.

The objective of this study is to develop a machine learning based predictive model from the available crash test data and use it for predicting injury metrics. In this study, a model was developed for predicting the head injury criterion, HIC15, using pre-test features (vehicle, test, occupant and restraint related). This problem was solved as a classification task, in which HIC15 with a threshold of 700 was divided into three classes i.e. low, medium and high. Crash test data was collected from the NHTSA database and was split into training and test datasets. Predictive models were developed from the training dataset using cross-validation while the test dataset was only used at the final step to evaluate the chosen predictive model. A logistic regression based predictive model was chosen as it demonstrated minimal overfitting and gave the highest F1 score (0.81) on the validation dataset. This chosen model gave a F1 score of 0.82 on the test (new/unseen) dataset.

This study investigates the use of machine learning methods for the selection of energy storage devices in military electrified vehicles. Powertrain electrification relies on proper selection of energy storage devices, in terms of chemistry, size, energy density, and power density, etc. Military vehicles largely vary in terms of weight, acceleration requirements, operating road environment, mission, etc. This study aims to assist the energy storage device selection for military vehicles using the data-drive approach. We use Machine Learning models to extract relationships between vehicle characteristics and requirements and the corresponding energy storage devices. After the training, the machine learning models can predict the ideal energy storage devices given the target vehicles design parameters as inputs. The predicted ideal energy storage devices can be treated as the initial design and modifications to that are made based on the validation results.

Rapid development of Laser Powder Bed Fusion (L-PBF) technology enables almost unconstrained design freedom for metallic parts and components in automotive industry. However, the mechanical properties of L-PBF alloys, AlSi10Mg for example, have shown significant differences when compared with their counterparts via conventional manufacturing process, due to the unique microstructure induced by extremely high heating and cooling rate. Therefore, microstructure informed material modeling approach is critical to fully unveil the process-structure-property correlation for such materials and enable the consideration of the effect of manufacturing during part design. Multi-scale material modeling approach, in which crystal plasticity finite element (CPFE) models were employed at the microscale, has been previously developed for L-PBF AlSi10Mg.

Engine starting vibration of hybrid vehicle with Toyota hybrid system has variations even in the same vehicle, and a large vibration that occurs rarely may cause stress to the passengers. The contribution analysis based on the vibration theory and statistical analysis has been done, but the primary factor of the rare large vibration has not been clarified because the number of factors is enormous. From this background, we apply machine learning that can reproduce multivariate and complicated relationships to analysis of variation factors of engine starting vibration. Variations in magnitude of the exciting force such as motor torque for starting the engine and in-cylinder pressure of the engine and timing of these forces are considered as factors of the variations. In addition, there are also nonlinear factors such as backlash of gears as a factor of variations.

There has been an ongoing need to increase the application of industrial robots to complete high-accuracy aerospace manufacturing and assembly tasks. However, the success of this is dependent on the ability of robotic systems to meet the tolerance requirements of the sector. Machine learning (ML) robot error compensation models have the potential to address this challenge. Artificial neural networks (ANNs) have been successful in increasing the accuracy of industrial robots. However, they have not always brought robotic accuracy within typical aerospace tolerances. Methods that have not yet been investigated to further optimize the use ANNs used in ML robot error compensation methods are presented in this paper. The focus of ML compensation methods has dominantly surrounded ANNs; there have been little to no investigations into other types of ML algorithms for their suitability as robot error compensation models.

The study showcases the strength of machine learning (ML) models in imitating the operation of an advanced engine concept - the gasoline compression ignition (GCI) - at low loads. The GCI engine is prone to exceeding the limits of criteria emissions at such loads, especially at the cold start when the catalyst is not activated. One proposition to accelerate catalyst light-off is using spark-ignition. This, however, adds an extra level of complexity in identifying an optimum operation point. The ML models can be a useful tool in guiding the engine calibration process. In this study, the ML models are trained on GCI engine experiments, covering different intake conditions, injection strategies, and spark settings. The models can predict seven engine performance parameters: fuel consumption, four engine-out emissions, exhaust temperature, and coefficient of variation (COV) in indicated mean effective pressure (IMEP).

This research evaluates the capability of data-science models to classify the combustion events in Cooperative Fuel Research Engine (CFR) operated under Homogeneous Charge Compression Ignition (HCCI) conditions. A total of 10,395 experimental data from the CFR engine at the University of Michigan (UM), operated under different input conditions for 15 different fuel blends, were utilized for the study. The combustion events happening under HCCI conditions in the CFR engine are classified into four different modes depending on the combustion phasing and cyclic variability (COVimep). The classes are; no ignition/high COVimep, operable combustion, high MPRR, and early CA50. Two machine learning (ML) models, K-nearest neighbors (KNN) and Support Vector Machines (SVM), are compared for their classification capabilities of combustion events. Seven conditions are used as the input features for the ML models viz.

The combustion system of a heavy-duty diesel engine operated in a gasoline compression ignition mode was optimized using a CFD-based response surface methodology and a machine learning genetic algorithm. One common dataset obtained from a CFD design of experiment campaign was used to construct response surfaces and train machine learning models. 128 designs were included in the campaign and were evaluated across three engine load conditions using the CONVERGE CFD solver. The design variables included piston bowl geometry, injector specifications, and swirl ratio, and the objective variables were fuel consumption, criteria emissions, and mechanical design constraints. In this study, the two approaches were extensively investigated and applied to a common dataset. The response surface-based approach utilized a combination of three modeling techniques to construct response surfaces to enhance the performance of predictions.

In an automotive braking system, Vacuum pump is used to generate vacuum in the vacuum servo or brake booster in order to enhance the safety and comfort to the driver. The vacuum pump operation in the braking system varies from conventional to electric vehicles. The vacuum pump is connected to the alternator shaft or CAM shaft in a conventional vehicle, operates continuously at engine speed and supplies continuous vacuum to the brake servo irrespective of vacuum requirement. To sustain continuous operation, these vacuum pumps are generally oil cooled. Whereas in electric vehicles, the use of a motor-driven vacuum pump is very much needed for vacuum generation as there is no engine present. Thus, with the assistance of an electronic control unit (ECU), the vacuum pump can be operated only when needed saving a significant amount of energy contributing to fuel economy and range improvement and emission reduction.

This paper describes a multi-disciplinary damage detection methodology that can aid in detecting and diagnosing a damage in a given structural system, not limited to the example of a rotating gear presented here. Damage detection is performed on the gear stress data corresponding to the steady state conditions. The normal and damage data are generated by a finite-difference solution of elastodynamic equations of velocity and stress in generalized coordinates1. The elastodynamic solution provides a knowledge of the stress distribution over the gear such as locations of stress extrema, which in turn can lead to an optimal placement of appropriate sensors over the gear to detect a potential damage. The damage detection is performed by a multi-function optimization that incorporates Tikhonov kernel regularization reinforced by an added Laplacian regularization term as used in semi-supervised machine learning. Damage is mimicked by reducing the rigidity of one of the gear teeth.

Despite the increasing number of electrified vehicles the transportation system still largely depends on the use of fossil fuels. One way to more rapidly reduce the dependency on fossil fuels in transport is to replace them with biofuels. Evaluating the potential of different biofuels in different applications requires knowledge of their physicochemical properties. In chemistry, message passing neural networks (MPNNs) correlating the atoms and bonds of a molecule to properties have shown promising results in predicting the properties of individual chemical components. In this article a machine learning approach, developed from the message passing neural network called Chemprop, is evaluated for the prediction of multiple properties of organic molecules (containing carbon, nitrogen, oxygen and hydrogen). A novel approach using transfer learning based on estimated property values from theoretical estimation methods is applied.

Dynamic skip fire (DSF) has shown significant fuel economy improvements via reduction of pumping losses that generally affect throttled spark-ignition engines. For production readiness, DSF engines must meet regulations for on-board diagnostics (OBD-II), which require detection and monitoring of misfire in all passenger vehicles powered by an internal combustion engine. Numerous misfire detection methods found in the literature, such as those using peak crankshaft angular acceleration, are generally not suitable for DSF engines due to added complexity of skipping cylinders. Specifically, crankshaft acceleration traces may change abruptly as the firing sequence changes. This article presents a novel method for misfire detection in a DSF engine using machine learning and artificial neural networks. Two machine learning approaches are presented.

This paper discusses an emerging area of applying machine learning (ML) methods to augment traditional Computational Fluid Dynamics (CFD) simulations of road vehicle aerodynamics. ML methods have the potential to both reduce the computational effort to predict a new geometry or car condition and to explore a greater number of design parameters with the same computational budget. Similar to traditional CFD methods, there exists a broad range of approaches. In particular, the accuracy and computational efficiency of a CFD simulation vary greatly depending on the choice of turbulence model (DNS, LES, RANS) and the underlying spatial and temporal numerical discretizations. Similarly, the end-user must select the correct ML method depending on the use-case, the available input data, and the trade-off between accuracy and computational cost. In this paper, we showcase several case studies using various data-driven ML methods to highlight the promise of these approaches.

This paper describes the initial results of applying two machine-learning-based unsupervised anomaly detection algorithms, Orca and GritBot, to data from two rocket propulsion testbeds. The first testbed uses historical data from the Space Shuttle Main Engine. The second testbed uses data from an experimental rocket engine test stand located at NASA Stennis Space Center. The paper describes four candidate anomalies detected by the two algorithms.

Recommender systems guide a user to useful objects in a large space of possible options in a personalized way. In this paper, we study recommender systems for vehicles. Compared to previous research on recommender systems in other domains (e.g., movies or music), there are two major challenges associated with recommending vehicles. First, typical customers purchase fewer cars than movies or pieces of music. Thus, it is difficult to obtain rich information about a customer’s vehicle purchase history. Second, content information obtained about a customer (e.g., demographics, vehicle preferences, etc.) is also difficult to acquire during a relatively short stay in a dealership. To address these two challenges, we propose an interactive vehicle recommender system based a novel machine learning method that integrates decision trees and multi-armed bandits. Decision tree learning effectively selects important questions to ask the customer and encodes the customer's key preferences.

This paper presents a detailed overview of the advantages and benefits of using 2-D barcodes, called Data Matrix codes, on Wing Fastening System (WFS) Tooling. This project was conducted on, but not limited to, the 777 Wing Fastening System (GEMCOR) tooling including the drills, fingers, and button dies. This paper will show how using Data Matrix codes to identify tooling will: Eliminate excessive downtime due to the operator using the incorrect tooling for a given tool setup. Reduce the cost associated with panel rework due to the use of incorrect tooling. Reduce the cost associated with excessive tool inventory or last minute ordering to keep up with production needs. Track tool life information for each specific tool. Provide operators with an easy to use tool setup reference document. And provide the factory with the ability to trace panel damage or defects back to the specific machine and exact tooling used.

Methods of testing brake linings by sample machine have been evolved to a standard of consistency which enables the performance of a brake to be evaluated by consideration of the geometry of the system and the coefficient of friction of the lining. This situation does, however, only hold if the way in which the lining is tested bears a close relationship to the duty cycle employed on an actual brake. In this paper the correlation between brake performance and estimates based on scale testing shows that a classification can be employed to simplify the choice of replacement linings when a single material is employed. Duo servo brakes are least amenable to this technique because of their high sensitivity.

Through effective utilization of the basic principles outlined in this paper, ECM equipment can become commonplace in manufacturing plants of the future. The three elements of an ECM machine are established by the following requirements for electrochemical machining: 1. The tool (cathode) must be positioned accurately relative to the workpiece (anode) and advanced toward the workpiece at a constant rate. There must be no relative movement between the tool and workpiece except in the direction in which machining is to be done. 2. A constant high velocity flow of clean electrolyte must be maintained in the small gap between the two electrodes. 3. Sufficient capacity of d-c current must be provided at machining gap to maintain the desired metal removal rate.

This paper gives an overview of todays machine vision technology with specific emphasis on microcomputer-based image processing and its potential as a low cost machine vision system. Topics discussed are general hardware requirements, image enhancement and segmentation techniques for binary and gray level images, two dimensional shape analysis, and additional sensors to supplement two dimensional image information.