NASA has a long history of collecting and analyzing data (e.g. images, spectrometry) of a multitude of planetary bodies in the solar system obtained through remote sensing via an array of instrumentation on spacecraft. The nature of the observations is highly dependent on the scientific goals and objectives of the mission, but certain technologies and instruments have been flown on multiple missions. In this presentation, we will discuss several different types of instruments and provide examples of the types of measurements made for selected missions to highlight both the usefulness and limitations of remotely sensed data in understanding various aspects of planetary science.

Adaptive Cruise Control (ACC) is one of the common advanced driving features that aims to assist the driver from fatigue. A lot of models for ACC are available, such as Proportional, Integral, and Derivative (PID), Model Predictive Control (MPC), and reinforcement learning (RL). Recently a deep RL technique called deep deterministic policy gradient (DDPG) has been widely used for velocity control. DDPG is a mixture of deep learning (DL) and RL method, that uses neural networks (NN) to estimate its action. In this work, we utilize an attention DDPG model. The attention mechanism in DDPG increases the overall effectiveness of the model by reducing focus on the less important features. The network structure consists of one hidden layer with 30 neurons, which has been deployed in our previous work. The reward function has been designed to concentrate on overall safety and comfort. In this paper, we introduce a new criterion of weather conditions, which has not been used before in the literature. We trained our model with a publicly available dataset. For testing, we created simulated sensor data created in the Mississippi State University Autonomous Vehicular Simulator (MAVS). We introduced various scenarios within the simulation environment with differing weather conditions to assess the performance of the ACC model. In many cases, the sensors are weather-sensitive, which can affect the performance of the ACC system. The primary objective of this study is to evaluate the performance of the Attention DDPG-based ACC model under varying weather conditions. The results show that the agent can maintain safety across a range of weather conditions.

Magnetic sensing has applications in diverse technological domains including geology and geophysics, structural inspection, geospatial navigation, and vehicle detection and avoidance. At-range detection of large assets such as vehicles, specifically, can be executed using UAV-deployed magnetometers to safely determine the presence or identity of vehicles even if they are visually obscured. However, several challenges must be overcome and, in particular, the electromagnetic noise generated by the aircraft motors and other onboard electronics sets the signal-to-noise ratio and thereby limits the effective range of such an application.

Here, we present UAV integration with a compact optically pumped magnetometer (QuSpin QTFM) and characterize its performance in vehicle sensing from the drone platform. We first characterize magnetic noise produced by the drone rotors to understand the detection limits it imposes on SNR. We then demonstrate vehicle detection at standoff distances of 5 and 10m. We utilize a volumetric magnetic field background subtraction method in order to enhance vehicle detection, and we characterize the impact of UAV velocity on field detection.

Real-time collection of reliable sensor information for the computing elements of Autonomous Systems (ASs) is crucial for execution of tasks that protect human lives and strategic assets. Both Cyber-Physical Systems (CPS), and the Internet of Things (IoT), heavily rely on dependable sensor data to perform their functions. We propose the utilization of a sensor-based Physically Unclonable Function (PUF) to detect the health of the sensor itself; without imposing restrictions on its potential application for cryptographic authentication, identification and key generation in future endeavors. Two key considerations are emphasized: (1) in the absence of sensor malfunction, predictable responses are anticipated and (2) if the sensor is malfunctioning then there should be a noticeable shift in the responses; indicating the need for replacement or the disregard of sensor data information (which may prevent an unintended outcome that may otherwise occur). We conduct an examination of both intra and inter aspects of the accelerometer-PUF. Furthermore, a dedicated section is included to present the outcomes of health detection based on the sensor pairs’ responses. Additionally, we perform temperature testing to measure the reproducibility of the PUF under varying conditions; encompassing both cold and hot temperatures over the range −20°C < T > 50°C.

Detecting and localizing large obstacles, particularly trees in unpaved regions, holds significant importance in autonomy, navigation, and related fields. Equally crucial is the extraction of detailed physical information from sensor captures. Accurately estimating the physical parameters of trees, such as the tree diameter at breast height (DBH), is particularly valuable for commercial and research purposes, especially in forestry and ecological studies. This estimation also plays a pivotal role in navigational tasks within densely vegetated regions, where overcoming obstacles becomes essential to achieving objectives. Achieving the required accuracy often entails labor-intensive processes such as manual collection aiming, data segmentation, or shape-building through mapping. In this context, we propose an algorithm based on particle swarm optimization (PSO) assisted Hough Transformation (HT) for tree DBH estimation, utilizing solely the physical spatial information from an actively available LiDAR point cloud. As a point of comparison, a straightforward circular HT-based method is also implemented. Our proposed approach surpasses the base HT method, demonstrating superior performance with an average error of 5.60 cm and RMSE of 6.57 cm, all while maintaining low time costs. These results reveal promising implications for this research direction in real-world applications, particularly in push-through navigation scenarios.

This paper tackles the new utilization of acoustic tomography and synthetic data modeling to improve autonomous navigation systems, especially in environments where traditional navigation tools, such as GPS, are either restricted or absent. Thus, acoustic tomography, based on sound waves, as well as synthetic data modeling which creates unrealized but realistic navigation scenarios, seems to be a very useful way of enhancing accessible navigation in terms of reliability, precision, and adaptability. This research via a comprehensive literature review, review, and synthesis of current research articulates the promise of these technologies to transform navigation systems. It focuses on the approach in the integration of acoustic sensing with computational models in developing the system autonomy, covering the technological advancements and the problems. Besides, implications of the research to the whole range of applications from underwater exploration to urban autonomous vehicles specify the requirement of innovation and exploration development in this sphere. By providing a broad review and evaluation of the interactions of acoustic tomography with synthetic data modeling, this paper seeks to promote progress in autonomous navigation technologies, laying the groundwork for the future studies and engineering work in navigation through the world's most difficult environments.

Since 2015, the National Reconnaissance Office (NRO) has participated in the Space Science and Technology (S&T) Partnership Forum with the United States Space Force (USSF) and the National Aeronautics and Space Administration (NASA). One of the research topics is Space Autonomy, addressing the disparate requirements of these three agencies. However, there are common operational needs to improve mission performance, reduce mission costs, incorporate technology advances rapidly, increase reliability and resiliency, reduce risk, and adapt to anomalies and environmental hazards. While researching these requirements, the related question of autonomous system trustworthiness has come to the fore.

NRO autonomy research focuses on space systems engaging in behaviors in which human operators have confidence with some level of decision-making. Recent modeling has shown progress to overcome physical limitations such as time and the space environment, but the most significant advances has been in trust and trustworthiness from autonomous systems. NRO’s acquisition focus and the need to provide intelligence data in shorter timelines means that additional attention has been applied to document

In the rapidly evolving landscape of military technology, the demand for autonomous vehicles (AVs) is increasing in both public and private sectors. These autonomous systems promise many benefits including enhanced efficiency, safety, and flexibility. To meet this demand, development of autonomous vehicles that are resilient and versatile are essential to the transport and reconnaissance market. The sensory perception of autonomous vehicles of any kind is paramount to their ability to navigate and localize in their environment. Typically, the sensors used for localization and mapping include LIDAR, IMU, GPS, and radar. Each of these has inherent weaknesses that must be accounted for in a robust system. This paper presents quantified results of simulated perturbations, artificial noise models, and other sensor challenges on autonomous vehicle platforms. The research aims to establish a foundation for robust autonomous systems, accounting for sensor limitations, environmental noise, and defense against nefarious attacks.

Unmanned aerial vehicles (UAV) and satellites are becoming increasingly popular in business, government, and military applications. Both have unique use cases and value, but they have several overlapping use cases and features. Most notably they are both used for observation, such as the case of climate monitoring or surveying and mapping. Satellites also have uses in communication and navigation by broadcasting signals and enabling technology such as global positioning systems (GPS). UAVs have also been deployed by the militaries across the world for both reconnaissance and offensive capabilities. Each are electro-mechanical systems with a several important components that need to be reliable and high performance. Maximizing the return in value for these assets might mean improving their performance, reliability, or longevity. One emerging technology that has the promise to do this is the digital twin (DT). DTs utilize a combination of multi-domain modeling and extensive data collection for real-time model updates. This real time updating can be utilized for advanced simulation, improved control, and advanced condition monitoring. DTs are an ideal platform for applying to UAVs and satellites to maximize their capabilities and values. As will be demonstrated in this work, DTs have been demonstrated to provide value in improving control performance, orientation and position tracking, condition monitoring, and fault detection in UAVs and satellites. A case study and preliminary work on a CubeSat attitude adjustment device DT has been presented and examined to display benefits of the concept.

Sensors are relied upon to represent the physical state of autonomous systems that are controlled by AI. Accurate data from these sensors is required to avoid collisions and catastrophic events in actual situations. Redundant sensors are often used in mission critical sensors to mitigate both unexpected failures and cyberattacks. Redundant sensors double or triple the sensor system cost without giving the user information about which sensor is faulty in real-time.

We developed a method that exploits small sensor-to-sensor variations; thereby fingerprinting them. For this purpose, we designed circuitry measuring a TMCS1101A4 magnetic current sensor pair. The small differences between the two sensors and the measurement of the calibrated sensor are amplified and recorded in a database as the current sensed is varied over a range from 0 to 320 mA. Future measurements of the sensor pair may be compared to the stored table in real-time to check sensor accuracy.

Sensor failure can be modeled by placing the output of one sensor in an open-circuit condition. Measurements then specify the percentage accuracy that is maintained in real-time without the sensor reporting a fault.

Additionally, arrays of sensors are used to generate unique system identifiers that verify the authenticity of the source of the data. These identifiers are not fixed but can be regenerated by a challenge from the server making it more difficult for actors to falsify the sensor data. Larger arrays of sensors will be investigated to generate encryption keys and seeds for Post Quantum Cryptography (PQC).

This research presents a novel technique for image compression and transmission tailored for Vehicular Ad Hoc Networks (VANETs), emphasizing the adaptation of Modulation and Coding schemes (MCS) to suit compression ratio demands contingent on vehicle density around roadside units. Leveraging Discrete Wavelet Transform Low-Low subband (DWT LL), Singular Value Decomposition (SVD), and Linear Programming (LP), the proposed technique dynamically adjusts compression ratios based on vehicular presence. Consequently, it enables simultaneous modulation of MCS and image compression, uniquely customizing images for VANET applications. Compression ratios ranging from 47.01% to 74.1% are achieved, correlating with vehicle density near Roadside units, satisfying size constraints crucial for efficient Infrastructure to Vehicles (I2V) image message dissemination within a 300-meter radius accommodating up to 200 vehicles. This adaptive approach holds promise for delivering image-enhanced public emergency notifications within the transportation sector, offering a viable solution for real-time communication and enhancing situational awareness amidst dynamic vehicular environments.

The Department of Defense (DoD) recognizes the imperative to embrace artificial intelligence (AI) and autonomy to effectively navigate the challenges of modern warfare. Leveraging data and analytics, including AI/autonomy, remains paramount across various DoD missions encompassing command and control, intelligence, logistics, and fires. To capitalize on the nation's AI expertise, the DoD has established AI Hubs to expedite research and innovation and launched the Accelerating Data and AI (ADA) initiative, embedding digital teams within Combatant Commands to facilitate agile decision-making. Furthermore, the Responsible AI Strategy and Implementation Pathway, released in June 2022, underscores the DoD's structured approach to fostering responsible AI development. In his keynote address, the Deputy Under Secretary of Defense, Dr. David Honey, will offer insights into AI, autonomy, and other critical technologies, alongside detailing how the Research and Engineering (R&E) office is streamlining processes to facilitate smoother collaboration with companies interested in doing business with the Department. By prioritizing the mitigation of unintended bias in AI models and establishing robust data governance policies, the DoD aims to uphold AI ethical principles and ensure the safety, security, and trustworthiness of AI and autonomous systems deployed in defense operations.

Safety and robust operation of autonomous vehicles is a pertinent concern for both defense and also civilian systems. From self-driving cars to autonomous Navy vessels and Army vehicles, malfunctions can have devastating consequences, including losses of life and infrastructure. Autonomous ground vehicles use a variety of sensors to image their environment: passive sensors such as RGB and thermal cameras, and active sensors such as RGBD, LIDAR, radar, and sonar. These sensors are either used alone or fused to accomplish the basic mobile autonomy tasks: obstacle avoidance, localization, mapping, and, subsequently, path-planning. In this paper, we will provide a qualitative and quantitative analysis of the limitations of ROS mapping algorithms when depth sensors—LIDAR and RGBD—are degraded or obscured, e.g., by dust, heavy rain, snow, or other noise. Aspects that will be investigated include form of the degradation and effect on the autonomous operations. This work will summarize the limitations of publicly available ROS mapping algorithms when depth sensors are degraded or obscured. Furthermore, this work will lay a foundation for developing robust autonomy algorithms that are resilient to possible degraded or obscured sensors.

Lane centering is a significant feature in the automotive industry and an important feature for advanced or autonomous vehicles, providing assistance to help drivers stay in their lane. The objective of this work is to utilize camera images which are processed by a lane-centering algorithm to make steering decisions. A model with a convolutional neural network (CNN) algorithm is used and simulated datasets are trained and tested. The goal is to make the program learn how to steer the vehicle autonomously and achieve end-to-end learning for steering command generation. The CNN can map raw pixels from cameras directly to steering commands without any intermediate feature engineering. A model is utilized which is created by NVIDIA researchers called the NVIDIA PilotNet. This network comprises five convolutional layers for feature extraction, followed by three fully connected layers for predicting the steering commands. The model is trained using two different sets of data to see how well the model performs with different types of data. The first set comes from Udacity’s Self-Driving Car Nanodegree Program, which uses their open-source Vehicle Simulation. Secondly, a dataset from the Mississippi State University Autonomous Vehicular Simulator (MAVS) is used. The training process involves reducing the error between predicted steering angles and the actual steering commands logged by the car. After the model has been trained, it is implemented for testing the car’s autonomous capabilities within the Self Driving Car Nanodegree Program Simulation. In this mode, the car demonstrates the ability to effectively track and navigate along the road lanes.

Digital twin technology can play a significant role in multiple robots’ navigation by providing a virtual representation of the physical environment, robots, and their interactions. This high detail simulation can allow efficient and accurate navigation in difficult sc enarios wh ile en abling co st eff ective rob ot sol utions. In thi s res earch a multi-robot navigation system is developed using a reinforcement learning augmented RRT*-Smart algorithm. This proposed framework attempts to introduce a more efficient solution for navigating a partially known, static environment, while making use of the strengths of a centralized multi-robot system. The RRT*-Smart algorithm will be used to generate paths inside of the digital twin simulation consisting of both pre-gathered environment data as well as the LiDAR data received from the robots during operation. The various parameters of the RRT*- Smart algorithm will be tuned for each situation by use of reinforcement learning to allow for more adaptability. The effectiveness of this system is tested by use of a real-life simulation, with robots navigating their paths by use of the Dynamic Window Approach. The LiDAR data received from each robot is used to both avoid obstacles during navigation, as well as update the digital-twin map used by the RRT*-SMART path planner.

We integrate advanced computer vision and Vehicle-to-Infrastructure (V2I) communication systems for effective intersection navigation. In the first phase, the YOLOv8 deep learning model is employed to accurately detect traffic lights, with specialized training on the S2TLD Dataset for precision. Then we establish seamless V2I communication in MAVS Simulation, allowing vehicles to receive Signal Phase and Timing (SPaT) messages from traffic lights, enabling autonomous adjustment of speed and behavior. Simulating the scenarios in a high-fidelity automotive simulator demonstrates accurate traffic light detection and timely phase information, promising safer and more efficient intersection navigation for autonomous vehicles.

While deep learning is a very popular subsection of machine learning and has been used for decades, an availability gap exists for both knowledge and datasets in unstructured environments or in-the-wild applications. Knowledge of mobility in these free environments is an important stepping-stone for both Department of Defense applications as well as industrial autonomy applications. A few datasets that exist for unstructured environments such as RELLIS-3D for robotics, RUGD for navigation, and GOOSE for perception; however, due to the limited selection of datasets for this type of environment, most deep learning algorithms have not been thoroughly tested on this scenario. In this article, we will implement multiple deep learning methods on an in-house dataset to evaluate performance. Specifically, this article investigates the performance of pretrained, publicly available YOLOv4, ResNet-50, and Single Shot Detector (SSD) models on detection of unknown object classes encountered in the wild for improved, safe, and reliable maneuverability with minimized impediment in unstructured environments. The models used are tested using a dataset developed in-house for unstructured environment studies, and their performance is assessed with multiple metrics. The data used in this experiment was collected by the United States Army Corps of Engineers Engineer Research and Development Center.

The paper researches a utilization of the Salp Swarm Algorithm (SSA), a bio-mimetic optimization technique, to improve path planning in Unmanned Ground Vehicles (UGVs). Because of the crucial role of the efficient and reliable path planning in the implementation of UGVs in such sectors as military, rescue operations, and agriculture, there is a need for algorithms that are capable of navigating complex environments. The concept of SSA, based on the natural swarming behavior of salps, represents a very promising approach that is characterized by the exploration and exploitation properties of the algorithm. This study evaluates the performance of the SSA relative to existing particle swarm optimization (PSO), in terms of path optimality, computational efficiency, and dynamic obstacle adaptability, through a number of simulated environments. Results show that the SSA has the potential to compete with the traditional algorithms in path efficiency and computational load. However, PSO shows slight superiority results compared to SSA. This study highlights the potency of bio-inspired algorithms, specifically the SSA, in enhancing the field of autonomous navigation for UGVs. It introduces new possibilities of practical application of SSA in real-life scenarios, demonstrating its scalability and resilience. The findings of this study make a contribution to the general discussion on the improvement of planning of autonomous routes and provide a possible way for more sustainable and effective UGV activities.