Unmanned Aerial Systems, hereafter referred to as UAS, are of great use in hazard events such as wildfire due to their ability to provide high-resolution video imagery over areas deemed too dangerous for manned aircraft and ground crews. This aerial perspective allows for identification of ground-based hazards such as spot fires and fire lines, and to communicate this information with fire fighting crews. Current technology relies on visual interpretation of UAS imagery, with little to no computer-assisted automatic detection. With the help of big labeled data and the significant increase of computing power, deep learning has seen great successes on object detection with fixed patterns, such as people and vehicles. However, little has been done for objects, such as spot fires, with amorphous and irregular shapes. Additional challenges arise when data are collected via UAS as high-resolution aerial images or videos; an ample solution must provide reasonable accuracy with low delays. In this paper, we examined 4K ( $3840\times 2160$ ) videos collected by UAS from a controlled burn and created a set of labeled video sets to be shared for public use. We introduce a coarse-to-fine framework to auto-detect wildfires that are sparse, small, and irregularly-shaped. The coarse detector adaptively selects the sub-regions that are likely to contain the objects of interest while the fine detector passes only the details of the sub-regions, rather than the entire 4K region, for further scrutiny. The proposed two-phase learning therefore greatly reduced time overhead and is capable of maintaining high accuracy. Compared against the real-time one-stage object backbone of YoloV3, the proposed methods improved the mean average precision(mAP) from $0.29$ to $0.67$ , with an average inference speed of 7.44 frames per second. Limitations and future work are discussed with regard to the design and the experiment results.

Connected and automated vehicles (CAVs) provide various valuable and advanced services to manufacturers, owners, mobility service providers, and transportation authorities. As a result, a large number of CAV applications have been proposed to improve the safety, mobility, and sustainability of the transportation system. With the increasing connectivity and automation, cybersecurity of the connected and automated transportation system (CATS) has raised attention to the transportation community in recent years. Vulnerabilities in CAVs can lead to breakdowns in the transportation system and compromise safety (e.g., causing crashes), performance (e.g., increasing congestion and reducing capacity), and fairness (e.g., vehicles fooling traffic signals). This paper presents our perspective on CATS cybersecurity via surveying recent pertinent studies focusing on the transportation system level, ranging from individual and multiple vehicles to the traffic network (including infrastructure). It also highlights threat analysis and risk assessment (TARA) tools and evaluation platforms, particularly for analyzing the CATS cybersecurity problem. Finally, this paper will provide valuable insights into developing secure CAV applications and investigating remaining open cybersecurity challenges that must be addressed.

Cooperative perception that integrates sensing capabilities from both infrastructure and vehicle perception sensors can greatly benefit the transportation system with respect to safety and data acquisition. In this study, we conduct a preliminary evaluation of such a system by integrating a portable lidar-based infrastructure detection system (namely, Traffic Scanner [TScan]) with a Society of Automotive Engineers (SAE) Level 4 connected and automated vehicle (CAV). Vehicle-to-everything (V2X) communication devices are installed on both the TScan and the CAV to enable real-time message transmission of detection results in the form of SAE J2735 basic safety messages. We validate the concept using a case study, which aims at improving CAV situation awareness and protecting vulnerable road user (VRU) safety. Field testing results demonstrate the safety benefits of cooperative perception from infrastructure sensors in detecting occluded VRUs and helping CAVs to plan safer (i.e., higher post-encroachment time) and smoother (i.e., lower deceleration rates) trajectories.

Prevalent solutions for Connected and Autonomous vehicle (CAV) mapping include high definition map (HD map) or real-time Simultaneous Localization and Mapping (SLAM). Both methods only rely on vehicle itself (onboard sensors or embedded maps) and can not adapt well to temporarily changed drivable areas such as work zones. Navigating CAVs in such areas heavily relies on how the vehicle defines drivable areas based on perception information. Difficulties in improving perception accuracy and ensuring the correct interpretation of perception results are challenging to the vehicle in these situations. This paper presents a prototype that introduces crowdsourcing trajectories information into the mapping process to enhance CAV's understanding on the drivable area and traffic rules. A Gaussian Mixture Model (GMM) is applied to construct the temporarily changed drivable area and occupancy grid map (OGM) based on crowdsourcing trajectories. The proposed method is compared with SLAM without any human driving information. Our method has adapted well with the downstream path planning and vehicle control module, and the CAV did not violate driving rule, which a pure SLAM method did not achieve.

Aiming at the imbalance between judgment accuracy and real-time performance of existing flame recognition models on devices with limited memory and computing capacity, a flame recognition model based on lightweight convolutional neural network YOLOV2 was designed. By using MobileNetV2 as the backbone feature extraction network of the target detection network YOLOV2, while modifying specific parameters, the amount of network parameters is greatly reduced. Experiments show that the accuracy of the detection model reaches 91.6%, the mAP index is 94.32%, the prediction time is 45 frames per second, and the model size is only 13MB, which effectively balances the accuracy and real-time performance of flame detection.