Safety-critical failures often have fatal consequences in aerospace control. Control systems on aircraft, therefore, must ensure the strict satisfaction of safety constraints, preferably with formal guarantees of safe behavior. This paper establishes the safety-critical control of fixed-wing aircraft in collision avoidance and geofencing tasks. A control framework is developed wherein a run-time assurance (RTA) system modulates the nominal flight controller of the aircraft whenever necessary to prevent it from colliding with other aircraft or crossing a boundary (geofence) in space. The RTA is formulated as a safety filter using control barrier functions (CBFs) with formal guarantees of safe behavior. CBFs are constructed and compared for a nonlinear kinematic fixed-wing aircraft model. The proposed CBF-based controllers showcase the capability of safely executing simultaneous collision avoidance and geofencing, as demonstrated by simulations on the kinematic model and a high-fidelity dynamical model.

Recent advancements in reinforcement learning techniques demonstrate an ability to make decisions in high dimen-sional state spaces and complex real-time strategy games. In contrast to supervised learning which features large data sets, there are relatively few existing environments for training rein-forcement learning agents. In addition, small differences in re-wards or action spaces can drastically change the difficulty and results of the training environments. Benchmarks seek to tackle both of these challenges by creating common environments, in the form of “Gyms” to train and compare reinforcement learning techniques, approaches, and algorithms. Many gyms, such as the classical control and Atari games environments, have become standard in new research on reinforcement learning. Researchers can easily compare and benchmark competing so-lutions across publications on these universal baselines enabling rapid innovation and collaboration. However, there are currently no standard set of environments for aerospace problems, and many of the gyms in the literature do not include safety con-straints or run time assurance systems that intervene when the reinforcement learning agent violates safety constraints. This manuscript describes the development of the Aerospace SafeRL Framework and accompanying Aerospace SafeRL Benchmarks that include interactive environments, safety constraints, soft-ware interfaces for run time assurance safety monitors with base implementations, and an initial set of baseline solutions. This initial set of scenarios introduces simple RL environments that expose the kinds of motion patterns, dynamics, and safety constraints encountered in air and space problems in 2D and 3D. This manuscript also describes standardized evaluation metrics for these environments to provide a consistent performance measurement with aerospace relevance. These benchmarks pro-vide a structured foundation for future reinforcement learning algorithms, run time assurance designs, and neural network verification techniques for the aerospace domain.

This paper presents an investigation of the effectiveness of the connected learning and integrated course knowledge (CLICK) approach. The CLICK approach aims to integrate the knowledge across the industrial engineering (IE) curriculum by leveraging immersive technology, i.e., 3D simulation and virtual reality (VR). The effectiveness of the CLICK approach is measured by its impact on students’ motivation, engineering identity, and learning outcomes. In this work, a virtual system that simulates a manufacturing assembly system was developed and used in an operations research (OR) course. The virtual system includes data collection tasks and exercises to calculate statistics that are taught in a probability and statistics course, and inventory and queueing theories concepts that are taught in an operations research course. The virtual system (CLICK learning module) is used to teach inventory and queueing theory concepts. Due to COVID-19 and the sudden shift to remote learning, the research team faced challenges including limitations in performing in-person experiments on campus as well as the potential risk of spreading the disease when VR headsets are used by several people. To alleviate some of the challenges, the researchers built the virtual system in simulation software, i.e., Simio, to provide more flexibility and scalability. The virtual system can be run on a regular personal computer without the need for a VR-ready computer and VR headsets. Yet, the virtual system can be run on an Oculus VR headset if the student prefers to do so. The study involves two groups: Control and intervention groups. The control group is represented by the students who are taught traditionally while the intervention group is represented by the students who are taught with the aid of the CLICK learning module. The results of this study compared the groups in terms of students’ motivation, and engineering identity. The learning outcomes were assessed using a self-assessment instrument and the student's grades in the learning module. The data of the control and intervention groups were collected at Penn State Behrend in Fall 2019, and Fall 2020 semesters, respectively. The groups were not statistically significantly different for motivation and Engineering Identity, however, the resulted motivation and Engineering Identity scores for the intervention group were not worse than the control group considering the shift to remote learning setting. The students showed good learning outcomes when the CLICK learning module was used. The grades were positively correlated to the motivation and Engineering Identity scores.

This paper presents the results and findings of the connected learning and integrated course knowledge (CLICK) approach. The CLICK approach aims to provide an integrative learning experience by leveraging virtual reality (VR) and 3D simulation technology. Integrative learning is described as the process of creating connections between concepts (skill and knowledge) from different resources and experiences, linking theory and practice, and using a variation of platforms to help students' understandings. With this approach, the integration is achieved by using a virtual system that mimics a real-life manufacturing or service system. VR and 3D simulation technology are chosen because they enhance visualization, interaction, and collaboration which makes them suitable for educational settings. In addition, immersive technologies provide the sense of being part of the environment. They are effective educational tools because they give students the ability to interact with objects and space in real-time compared to traditional distance, time, or safety constraints. Virtual systems can be designed and created to provide an integrative learning environment via a theme that connects and transfers the knowledge across a curriculum. The paper will focus on the results of the project from two perspectives: technological and educational. The technological perspective will describe the research efforts of automatically generating virtual environments using the reinforcement learning (RL) approach while the educational perspective will summarize the results on the effectiveness of the CLICK approach on students’ motivation, engineering identity, and learning outcomes.

This work presents a deep reinforcement learning (DRL) approach for procedural content generation (PCG) to automatically generate three-dimensional (3D) virtual environments that users can interact with. The primary objective of PCG methods is to algorithmically generate new content in order to improve user experience. Researchers have started exploring the use of machine learning (ML) methods to generate content. However, these approaches frequently implement supervised ML algorithms that require initial datasets to train their generative models. In contrast, RL algorithms do not require training data to be collected a priori since they take advantage of simulation to train their models. Considering the advantages of RL algorithms, this work presents a method that generates new 3D virtual environments by training an RL agent using a 3D simulation platform. This work extends the authors’ previous work and presents the results of a case study that supports the capability of the proposed method to generate new 3D virtual environments. The ability to automatically generate new content has the potential to maintain users’ engagement in a wide variety of applications such as virtual reality applications for education and training, and engineering conceptual design.

The authors present a generative adversarial network (GAN) model that demonstrates how to generate 3D models in their native format so that they can be either evaluated using complex simulation environments or realized using methods such as additive manufacturing. Once initially trained, the GAN can create additional training data itself by generating new designs, evaluating them in a physics-based virtual environment, and adding the high performing ones to the training set. A case study involving a GAN model that is initially trained on 4045 3D aircraft models is used for demonstration, where a training data set that has been updated with GAN-generated and evaluated designs results in enhanced model generation, in both the geometric feasibility and performance of the designs. Z-tests on the performance scores of the generated aircraft models indicate a statistically significant improvement in the functionality of the generated models after three iterations of the training-evaluation process. In the case study, a number of techniques are explored to structure the generate-evaluate process in order to balance the need to generate feasible designs with the need for innovative designs.

In this work, a Deep Reinforcement Learning (RL) approach is proposed for Procedural Content Generation (PCG) that seeks to automate the generation of multiple related virtual reality (VR) environments for enhanced personalized learning. This allows for the user to be exposed to multiple virtual scenarios that demonstrate a consistent theme, which is especially valuable in an educational context. RL approaches to PCG offer the advantage of not requiring training data, as opposed to other PCG approaches that employ supervised learning approaches. This work advances the state of the art in RL-based PCG by demonstrating the ability to generate a diversity of contexts in order to teach the same underlying concept. A case study is presented that demonstrates the feasibility of the proposed RL-based PCG method using examples of probability distributions in both manufacturing facility and grocery store virtual environments. The method demonstrated in this paper has the potential to enable the automatic generation of a variety of virtual environments that are connected by a common concept or theme.

The objective of this work is to present an initial investigation of the impact the Connected Learning and Integrated Course Knowledge (CLICK) approach has had on students’ motivation, engineering identity, and learning outcomes. CLICK is an approach that leverages Virtual Reality (VR) technology to provide an integrative learning experience in the Industrial Engineering (IE) curriculum. To achieve this integration, the approach aims to leverage VR learning modules to simulate a variety of systems. The VR learning modules offer an immersive experience and provide the context for real-life applications. The virtual simulated system represents a theme to transfer the system concepts and knowledge across multiple IE courses as well as connect the experience with real-world applications. The CLICK approach has the combined effect of immersion and learning-by-doing benefits. In this work, VR learning modules are developed for a simulated manufacturing system. The modules teach the concepts of measures of location and dispersion, which are used in an introductory probability course within the IE curriculum. This work presents the initial results of comparing the motivation, engineering identity, and knowledge gain between a control and an intervention group (i.e., traditional vs. CLICK teaching groups). The CLICK approach group showed greater motivation compared to a traditional teaching group. However, there were no effects on engineering identity and knowledge gain. Nevertheless, it is hypothesized that the VR learning modules will have a positive impact on the students’ motivation, engineering identity, and knowledge gain over the long run and when used across the curriculum. Moreover, IE instructors interested in providing an immersive and integrative learning experience to their students could leverage the VR learning modules developed for this project.

We present a policy and process framework for secure environments for productive data science research projects at scale, by combining prevailing data security threat and risk profiles into five sensitivity tiers, and, at each tier, specifying recommended policies for data classification, data ingress, software ingress, data egress, user access, user device control, and analysis environments. By presenting design patterns for security choices for each tier, and using software defined infrastructure so that a different, independent, secure research environment can be instantiated for each project appropriate to its classification, we hope to maximise researcher productivity and minimise risk, allowing research organisations to operate with confidence.

Stable, anatomical fixation of acetabular fractures gives the best chance of successful outcome, while penetration of the acetabular articular surface with screws is associated with poor outcomes. Spring plates are an alternative to interfragmentary lag screws when penetration is a concern. A mechanical study comparing fracture stability and construct stiffness of three fixation methods for posterior wall acetabular fractures with transverse comminutions was performed. The three fixation methods tested were a posterior wall rim plate, a posterior wall buttress plate with separate lag screws and a posterior wall plate with two spring plates. Nine samples were tested, three for each fixation method. Two-dimensional motion analysis was used to measure fracture fragment displacement and construct stiffness. After two 6000 cycle-loading protocols, to a maximum 1.5 kN, the mean fracture displacement was 0.154 mm for the rim plate model, 0.326 mm for the buttress plate and 0.254 mm for the spring plate model. Mean maximum displacement was significantly less for the rim plate fixation than the buttress plate (p = 0.015) and spring plate fixation (p = 0.02). The rim plate was the stiffest construct 10,962 N/mm, followed by the spring plate model 5637 N/mm and the buttress plate model 4882 N/mm. Based on data obtained in this study, where possible a rim plate with interfragmentary lag screws should be used for isolated posterior wall fractures as this is the stiffest and most stable construct. When this method is not possible, spring plate fixation is a safe and a superior alternative to a posterior buttress plate method.

Small branching veins that arise from the venous outflow of surgical arterial-venous fistulas (AVFs) are frequently seen during fistulograms performed to evaluate for poorly functioning AVFs. It is hypothesized that the presence of escape veins can decrease the performance of native AVFs during hemodialysis by diverting flow. Though interventional methods for exclusion of escape veins are effective, the mechanism of disruption these small branching vessels cause on flow through AVFs is unknown. Furthermore, an objective method for identifying escape veins that cause significantly diminished venous flow has not been defined. The following describes the detrimental nature of escape veins using tenants of physics and electrical circuitry. Subsequently, the proceeding study shows the identification of small branching escape veins in patients during fistulography. Intravascular pressure measurements were obtained proximal and distal to the ostium of the offending collaterals in these patients. Escape veins causing a pressure gradient of at least 5 mmHg were treated, and pressure measurements were repeated following intervention. The patients were entered into a database and hemodialysis blood flow rates were monitored to determine if escape vein intervention increased AVF performance.

A transjugular intrahepatic portosystemic shunt is the standard of care for complications of portal hypertension, such as variceal bleeding, refractory ascites, and hepatic hydrothorax. Hepatic encephalopathy, hepatic insufficiency, and right heart failure are the major complications after shunt creation. If medical management is unsuccessful, the interventionalist is consulted to close/revise the shunt. Closure of the shunt results in a dangerously abrupt increase in portal pressure, increasing risk for life-threatening variceal bleeding. Methods for revising these shunts are reported, which result in coarse adjustments in shunt diameter, causing rapid changes in portosystemic gradients. Our method for shunt revision utilizes carefully sized covered stents deployed in a controlled "onion skin" fashion to produce a narrowing within the hepatic venous limb of the shunt to precisely calibrate the desired portosystemic gradient.

In this paper, we present a method that uses a physics based virtual environment to evaluate the feasibility of neural network-based generated designs. Deep learning models rely on large training data sets that are used for training. These training data sets are typically validated by human designers that have a conceptual understanding of the problem being solved. However, the requirement of human training data severely constrains the size and availability of training data for computer generated models due to the manual process of either creating or labeling such data sets. Furthermore, there may be misclassification errors that result from human labeling. To mitigate these challenges, we present a physics-based simulation environment that helps users discover correlations between the form of a generated design and the physical constraints that relate to its function. We hypothesize that training data that includes machine validated designs from a physics-based virtual environment will increase the probability of generative models creating functionally-feasible design concepts. A case study involving a generative model that is trained on over 70,000 human 2D boat sketches is used to test the hypothesis. Knowledge gained from testing this hypothesis will provide human designers with insights into the importance of training data in the resulting design solutions generated by deep neural networks.

The differential diagnosis of idiopathic Parkinson's Disease (iPD) from atypical parkinsonian syndromes can be very difficult at the early stages of these diseases. Trancranial Ultrasound Imaging (TCUI) of the Substantia Nigra (SN) is one method that has been shown to aide in this early differential diagnosis. TCUI is done to detect hyperechogenicity in the SN, which is defined as an echogenic area above a threshold size of 0.2 cm 2 . Because B-mode ultrasound images are often noisy, determining the size of the echogenic area can be difficult. Harmonic imaging using a Third-Order Volterra (ToVF) filter is one solution that has been successful in filtering out the noise in these images, allowing a more reliable diagnosis. In this paper, we show that regularization methods such as Truncated Singular Value Decomposition (TSVD) and Tikhonov method can be used to solve for the Volterra Filter's coefficient much more quickly than least mean square (LMS) methods studied previously without sacrificing image quality. This finding has implications in terms of the Volterra Filter's viability for use in real-time harmonic imaging applications.

This work presents a deep neural network method for approximating the performance of generated design concepts. This deep learning meta-modeling approach minimizes the need for costly simulations that test for design concept feasibility by discovering the visual features of a design that correlated to good and bad performance. These form-function relationships are discovered by simply observing the pixels of images of many candidate designs and their corresponding performance in a simulation environment. As opposed to existing metamodeling techniques, this evaluation is agnostic to the simulation environment and applicable to any design space in which form and function are closely linked. A case study is presented in which 2D sketches of boats generated from a deep generative model are evaluated in a simulation environment based on their ability to travel through water without sinking as well as their speed of travel. It is shown through simulation that 57.5% of the designs, which are validated according to their form during the generation process, fail in their intended function. Additionally, the trained VNN is able to classify designs it has never seen before as successful or failing with an accuracy of 86.6% and an F1-Score of 0.806.

A 65-year-old woman from India presented with a four-month history of demyelinating motor neuropathy, pancytopenia, including severe absolute neutropenia, and recently evolving CNS lesions. Four bone marrow (BM) biopsies performed and interpreted at the outside institution showed hemophagocytosis in the absence of obvious neoplastic or infectious disorders. A trial of steroids produced a marginal and unsustained improvement. Repeat BM biopsy at admission showed numerous atypical lymphoid cells (Image 1, panel A) demonstrating a mature clonal B-cell phenotype on concurrent flow cytometry (positive for CD19, CD20, CD38[variable], surface Ig kappa and equivocal CD10), as well as hemophagocytosis (Image 1, panel B). While inconspicuous on the H&E (Image 1, panel C), these atypical cells were identified as exceedingly rare BSAP (PAX5)-positive B cells [brown] within vessels outlined by CD34-positive endothelial cells [red] (Image 1, panel D)] in the BM biopsy at presentation, confirming a diagnosis of intravascular large B-cell lymphoma (IVLBCL). The patient passed away prior to therapeutic intervention. Wright Giemsa-stained bone marrow aspirate smears performed at admission show atypical lymphocytes (panel A) and hemophagocytosis (panel B). The atypical lymphocytes are inconspicuous on the H&E-stained bone marrow biopsy (panel C), but are highlighted as rare BSAP-positive B cells (brown) within the CD34-positive vessels (red) on immunohistochemical double stain (panel D) In Asian patients IVLBCL frequently presents with microscopically subtle BM involvement (75%) and, as in our case, occasionally associated hemophagocytic lymphohistiocytosis; neurologic symptoms and cutaneous lesions are less common than in Western countries. A BM biopsy with immunohistochemical stain for B-cell antigens should always be performed in patients with this constellation of symptoms. While little is known about the molecular landscape of IVLBCL, NGS-based molecular analysis in our case revealed NOTCH2 p.P2419fs* at the time of progression, a molecular abnormality found in 4%-8% of DLBCL. Nothing to report.

The management of lung cancer is highly dependent on surgical resection. There are different surgical managements that are utilized on a patient to patient basis. Small lung nodules are particularly difficult to resect and have a higher postoperative complication rate. Video-assisted thoracoscopic surgery is the preferred method of surgery over open thoracotomy, but requires the preoperative percutaneous placement of a marker to help the surgeon identify the nodule once conducting surgical resection. There are various methods to place percutaneous markers, but have reported complications. The following case will present the novel placement of a hydrogel tagging system on a small pulmonary nodule, demonstrating superiority to other methods of percutaneous marker placement.

A 69-year-old obese man was involved in a high-speed head-on motor vehicle collision. He was tachycardic and normotensive on arrival. He subsequently developed hemodynamic instability requiring blood transfusion. On examination he had bilateral pneumothoraces, an anterior-posterior compression (APC) pelvic fracture, an open wound at the left groin, and gross hematuria after Foley catheter placement.CT imaging revealed hemoperitoneum, right hepatic lobe grade II lacerations, splenic laceration, mesenteric root injury with extravasated contrast, intraperitoneal and extraperitoneal bladder rupture, bilateral ureteral injuries at the level of the pelvic inlet (see figure 1), APC pelvic fracture, bilateral rib fractures, pneumothoraces, and pulmonary contusions.Figure 1CT of the abdomen and pelvis with cystogram. Delayed images demonstrating accumulation of contrast in the retroperitoneum arising from the right and left ureter at the level of the pelvic brim. Extraluminal contrast from the intraperitoneal bladder injury is also identified.He underwent emergent exploratory laparotomy. Exploration confirmed the injuries noted on the CT scan. Hepatorrhaphy with abdominal and preperitoneal pelvic packing was performed. A large anterior bladder wall injury was visualized. Neither ureteral orifice was seen. The right ureter was completely transected at the level of the pelvic brim. The left ureter was decompressed and the full extent of its injury was not determined; however, the bladder injury left concern for a distal avulsion. The patient continued to be in shock.WHAT WOULD YOU DO?:Reconstruct the urinary bladder and reimplant bilateral ureters.Ligate the ureter and prepare for pelvic embolization and nephrostomy tubes.Continue to explore looking for the full extent of the left ureter.Externalize the ureters to the abdominal wall with the open abdomen.

During the early stage of the disease, idiopathic Parkinson's Disease can be very difficult to differentiate from atypical parkinsonian syndromes. Hyperechogenicity in the substantia nigra is one marker that has been shown to help make this differential diagnosis, and Transcranial Ultrasound Imaging is the preferred method for detecting SN hyperechogenicity. Hyperechogenicity is defined as an echogenic area larger than 0.2cm(2). However, B-mode imaging often contains enough noise that the boundary may not be clear, thus making this diagnosis much more difficult. Harmonic imaging using a Third-Order Volterra filter is one solution that has been shown to be successful in filtering out the noise in these images. In this paper we show that regularization methods such as the Truncated Singular Value Decomposition and Damped Singular Value Decomposition can be used to solve for the Volterra Filter's coefficients much more quickly than adaptive Least Mean Squared methods without sacrifice in image quality. These findings have significant implications for the viability of using the Volterra Filter in real-time applications.

Differentiating between the early stages of Parkinson's Disease (PD) and other diseases with parkinsonian symptoms is difficult from analyzing motor degeneration symptoms alone. For this reason, a commonly used diagnostic marker for PD is the hyperechogenicity of the Substantia Nigra (SN), which can help to make an early differential diagnosis of PD. Current practice for determining if an image displays hyper-echogenicty relies on clinician experience heavily because of the difficulty of discerning features in standard B-mode imaging. Harmonic imaging has been studied extensively, and while it does improve the image quality, it suffers from spectral overlap with the noisy fundamental component. Our approach uses an adaptive Third Order Volterra Filter (ToVF), which avoids this problem by completely separating an image into its linear, quadratic, and cubic components with no overlap. One of the standard implementations of the ToVF is through an adaptive Recursive Least Squares (RLS) algorithm. This paper examines two algorithms developed through applying an ℓ 0 constraint on the standard RLS cost function. The two algorithms approximate this cost function in different ways, one using a Slow Time Varying (STV) approximation and the other using a Taylor Series Expansion (TSE) approximation. Theoretically the ℓ 0 constraint will shorten the number of iterations to reach steady state without sacrificing image quality. Our results confirm that these theoretical results hold on an in vivo application.