Abstract: An internal combustion engine is described. Controlling the internal combustion engine includes gathering engine operating data during steady-state engine operation, including gathering a first dataset associated with a cylinder air charge during steady-state operation of the engine in the PVO state and gathering a second dataset associated with a cylinder air charge during steady-state operation of the engine in the NVO state. An optimization routine is executed to determine a first subset of parameters associated with a first relationship for a cylinder air charge model based upon the second dataset. The optimization routine is also executed to determine a second subset of parameters associated with a second relationship for the cylinder air charge model based upon the first dataset. A cylinder air charge is determined in real-time during engine operation based upon the cylinder air charge model and the first and second subsets of parameters.

Abstract: A vehicle is described, and includes a drivetrain system including an internal combustion engine and a geartrain, an accessory device, and a route monitoring system. A controller includes an instruction set to monitor operation of the accessory device, and a request to operate the accessory device. When the request to operate the accessory device includes a request to activate the accessory device based upon not exceeding a limit for an operating parameter of the accessory device, the instruction set executes to determine a horizon for the vehicle based upon the projected path for the vehicle, determine a projected road load and a speed constraint based upon the horizon for the vehicle, and determine a projected axle torque command based upon the projected road load. A first routine determines a first command for operating the accessory device that minimizes fuel consumption and meets the projected axle torque command.

Abstract: A vehicle engine system includes a combustion engine configured to provide a propulsion torque to satisfy a propulsion demand in response to a driver input. The engine system also includes a controller programmed to receive data indicative of at least one upcoming driving event, and issue a command to impart a predetermined velocity profile based on the upcoming driving event. The predetermined velocity profile is arranged to optimize a performance attribute of the combustion engine and preempt the driver input.

Abstract: A method for generating energy-optimized travel routes with a vehicle navigation system includes generating candidate travel routes between a route origin and one or more route destinations, and then dividing each candidate travel route into a plurality of route segments. The method includes estimating expected travel speeds along each segment using cloud information and calculating an expected energy efficiency over each of the candidate travel routes using one or more vehicle-specific energy efficiency models. The travel routes are displayed via the navigation system, including a trace of the energy-optimized travel routes and an expected or relative energy efficiency along the energy-optimized travel routes. A vehicle includes the navigation system and a powertrain. A powertrain controller may control vehicle speed over a selected route to maintain an optimally energy-efficient speed.

Abstract: A system according to the principles of the present disclosure includes a pedal position prediction module and an engine actuator control module. The pedal position prediction module predicts a pedal position at a future time based on driver behavior and vehicle driving conditions. The pedal position includes at least one of an accelerator pedal position and a brake pedal position. The engine actuator control module controls an actuator of an engine based on the predicted pedal position.

Abstract: Techniques for image captioning with weak supervision are described herein. In implementations, weak supervision data regarding a target image is obtained and utilized to provide detail information that supplements global image concepts derived for image captioning. Weak supervision data refers to noisy data that is not closely curated and may include errors. Given a target image, weak supervision data for visually similar images may be collected from sources of weakly annotated images, such as online social networks. Generally, images posted online include “weak” annotations in the form of tags, titles, labels, and short descriptions added by users. Weak supervision data for the target image is generated by extracting keywords for visually similar images discovered in the different sources. The keywords included in the weak supervision data are then employed to modulate weights applied for probabilistic classifications during image captioning analysis.

Abstract: An internal combustion engine is configured to operate in a homogeneous-charge compression-ignition combustion mode. Operation of the engine includes determining a combustion pressure parameter for each cylinder. Fueling for each cylinder is controlled responsive to a target state for the combustion pressure parameter for the corresponding cylinder. An end-of-injection timing and a corresponding spark ignition timing for each cylinder are controlled responsive to a target mass-burn-fraction point for an engine operating point.

Abstract: Systems and methods are provided for material processing. An example apparatus includes a process-kit component containing a first groove and a second groove. The first groove and the second groove are disposed to form a pattern on a surface of the process-kit component. The process-kit component is configured to be placed into a chamber to reduce material deposition on one or more parts of the chamber during material processing.

Abstract: Techniques for image captioning with word vector representations are described. In implementations, instead of outputting results of caption analysis directly, the framework is adapted to output points in a semantic word vector space. These word vector representations reflect distance values in the context of the semantic word vector space. In this approach, words are mapped into a vector space and the results of caption analysis are expressed as points in the vector space that capture semantics between words. In the vector space, similar concepts with have small distance values. The word vectors are not tied to particular words or a single dictionary. A post-processing step is employed to map the points to words and convert the word vector representations to captions. Accordingly, conversion is delayed to a later stage in the process.

Abstract: An internal combustion engine includes an air charging system. A method to control the air charging system includes providing a desired operating target command for the air charging system, and monitoring operating parameters of the air charging system. An error between the desired operating target command for the air charging system and the corresponding one of said operating parameters of the air charging system is determined, and scheduled PID gains are determined based on the error utilizing a PID controller. An adaptive algorithm is applied to modify the scheduled PID gains, and a system control command for the air charging system is determined based upon the modified scheduled PID gains. The air charging system is controlled based upon the system control command for the air charging system.

Abstract: An antenna with swappable and selective radiation direction includes a first arm, a second arm electrically connected to the first arm, a third arm is electrically connected to the first arm, a first impedance tuning circuit coupled to the second arm for connecting the second arm to a ground or a first matching component according to a control signal, and a second impedance tuning circuit coupled to the third arm for connecting the third arm to the ground or a second matching component according to the control signal. By tuning impedance of the antenna, the antenna operates in a first mode corresponding to a first radiation direction or a second mode corresponding to a second radiation direction.

Abstract: An engine assembly includes a control module configured to receive a torque request and an engine configured to produce an output torque in response to the torque request. The control module includes a processor and tangible, non-transitory memory on which is recorded instructions for executing a method for supervisory model predictive control. The control module includes a multi-layered structure with an upper-level (“UL”) optimizer module configured to optimize at least one system-level objective and a lower-level (“LL”) tracking control module configured to maintain at least one tracking parameter. The multi-layered structure is characterized by a decoupled cost function such that the UL optimizer module minimizes an upper-level cost function (CFUL) and the LL tracking control module minimizes a lower-level cost function (CFLL). The system-level objective may include minimizing fuel consumption of the engine and the tracking parameter may include delivering the torque requested to engine.

Abstract: An engine assembly includes a control module configured to receive a torque request and an engine configured to produce an output torque in response to the torque request. The control module includes a processor and tangible, non-transitory memory on which is recorded instructions for executing a method for supervisory model predictive control. The control module includes a multi-layered structure with an upper-level (“UL”) optimizer module configured to optimize at least one system-level objective and a lower-level (“LL”) tracking control module configured to maintain at least one tracking parameter. The multi-layered structure is characterized by a decoupled cost function such that the UL optimizer module minimizes an upper-level cost function (CFUL) and the LL tracking control module minimizes a lower-level cost function (CFLL). The system-level objective may include minimizing fuel consumption of the engine and the tracking parameter may include delivering the torque requested to engine.

Abstract: The present application relates to a new high-speed rail rescue crane and a multifunctional high-speed rail rescue train, comprising a vehicle equipped with wheels; a concave floor disposed on middle of the vehicle floor; wherein a steel frame base is disposed on the concave floor; a cassette deck is set on both sides of the steel frame base and a V-type chute is set up on the cassette deck; a pedestal floor is disposed on the steel frame base, a sliding tenon is set on both sides of the pedestal floor, the sliding tenons is engaged with the V-type chute; an end of the pedestal floor is connected to a trip lever of a hydraulic cylinder fixed on the steel frame base; a rotary support is fixed on the pedestal floor, a crane is fixed on the pedestal floor; legs of the crane are disposed on the vehicle floor; and a slot engaged with a steel rail is disposed on root of the legs of the crane.

Abstract: Embodiments of the present invention provide an automated image tagging system that can predict a set of tags, along with relevance scores, that can be used for keyword-based image retrieval, image tag proposal, and image tag auto-completion based on user input. Initially, during training, a clustering technique is utilized to reduce cluster imbalance in the data that is input into a convolutional neural network (CNN) for training feature data. In embodiments, the clustering technique can also be utilized to compute data point similarity that can be utilized for tag propagation (to tag untagged images). During testing, a diversity based voting framework is utilized to overcome user tagging biases. In some embodiments, bigram re-weighting can down-weight a keyword that is likely to be part of a bigram based on a predicted tag set.

Abstract: Embodiments of the present invention provide an automated image tagging system that can predict a set of tags, along with relevance scores, that can be used for keyword-based image retrieval, image tag proposal, and image tag auto-completion based on user input. Initially, during training, a clustering technique is utilized to reduce cluster imbalance in the data that is input into a convolutional neural network (CNN) for training feature data. In embodiments, the clustering technique can also be utilized to compute data point similarity that can be utilized for tag propagation (to tag untagged images). During testing, a diversity based voting framework is utilized to overcome user tagging biases. In some embodiments, bigram re-weighting can down-weight a keyword that is likely to be part of a bigram based on a predicted tag set.

Abstract: Certain embodiments involve learning features of content items (e.g., images) based on web data and user behavior data. For example, a system determines latent factors from the content items based on data including a user's text query or keyword query for a content item and the user's interaction with the content items based on the query (e.g., a user's click on a content item resulting from a search using the text query). The system uses the latent factors to learn features of the content items. The system uses a previously learned feature of the content items for iterating the process of learning features of the content items to learn additional features of the content items, which improves the accuracy with which the system is used to learn other features of the content items.

Abstract: Embedding space for images with multiple text labels is described. In the embedding space both text labels and image regions are embedded. The text labels embedded describe semantic concepts that can be exhibited in image content. The embedding space is trained to semantically relate the embedded text labels so that labels like “sun” and “sunset” are more closely related than “sun” and “bird”. Training the embedding space also includes mapping representative images, having image content which exemplifies the semantic concepts, to respective text labels. Unlike conventional techniques that embed an entire training image into the embedding space for each text label associated with the training image, the techniques described herein process a training image to generate regions that correspond to the multiple text labels. The regions of the training image are then embedded into the training space in a manner that maps the regions to the corresponding text labels.

Abstract: Modeling semantic concepts in an embedding space as distributions is described. In the embedding space, both images and text labels are represented. The text labels describe semantic concepts that are exhibited in image content. In the embedding space, the semantic concepts described by the text labels are modeled as distributions. By using distributions, each semantic concept is modeled as a continuous cluster which can overlap other clusters that model other semantic concepts. For example, a distribution for the semantic concept “apple” can overlap distributions for the semantic concepts “fruit” and “tree” since can refer to both a fruit and a tree. In contrast to using distributions, conventionally configured visual-semantic embedding spaces represent a semantic concept as a single point. Thus, unlike these conventionally configured embedding spaces, the embedding spaces described herein are generated to model semantic concepts as distributions, such as Gaussian distributions, Gaussian mixtures, and so on.

Abstract: Techniques for image captioning with word vector representations are described. In implementations, instead of outputting results of caption analysis directly, the framework is adapted to output points in a semantic word vector space. These word vector representations reflect distance values in the context of the semantic word vector space. In this approach, words are mapped into a vector space and the results of caption analysis are expressed as points in the vector space that capture semantics between words. In the vector space, similar concepts with have small distance values. The word vectors are not tied to particular words or a single dictionary. A post-processing step is employed to map the points to words and convert the word vector representations to captions. Accordingly, conversion is delayed to a later stage in the process.