The effectiveness of naive optimization of the egress path for an active-shooter scenario.

There have been 130 mass shootings in the United States from 1982 to June, 2022 according to the Mother Jones database of active shooter events. In these critical scenarios, making the right decisions while evacuating can be the difference between life and death. However, emergency evacuation is intensely stressful, which along with lack of verifiable real-time information may lead to costly incorrect decisions. In this paper, we demonstrate the effectiveness of a non-homogeneous semi-Markov-Decision-Process (NHSMDP) based naive algorithm that relies on prior knowledge about the layout of a building and uses recurring updates of the shooter's location (based on automatic processing of images from a camera network) to provide an optimized egress plan for evacuees. While emergency evacuations due to fire and natural disasters are well researched, the novelty of this work is in the response to a threat that moves either purposefully or randomly through the building and in incorporating the ability for an evacuee to wait for danger to pass before beginning egress and during the process of evacuation. This ability to include sojourn times in the optimized scheme is due to the NHSMDP formulation and is a notable augmentation to the current state-of-the-art. We show that following this algorithm can reduce casualties by 56% and the time spent by evacuees in the shooter's line of sight by 52% compared to an intuitive natural response guided by expert advice.

Automated vehicle data pipeline for accident reconstruction: New insights from LiDAR, camera, and radar data.

As automated vehicles are deployed across the world, it has become critically important to understand how these vehicles interact with each other, as well as with other conventional vehicles on the road. One such method to achieve a deeper understanding of the safety implications for Automated Vehicles (AVs) is to analyze instances where AVs were involved in crashes. Unfortunately, this poses a steep challenge to crash-scene investigators. It is virtually impossible to fully understand the factors that contributed to an AV involved crash without taking into account the vehicle's perception and decision making. Furthermore, there is a tremendous amount of data that could provide insight into these crashes that is currently unused, as it also requires a deep understanding of the sensors and data management of the vehicle. To alleviate these problems, we propose a data pipeline that takes raw data from all on-board AV sensors such as LiDAR, radar, cameras, IMU's, and GPS's. We process this data into visual results that can be analyzed by crash scene investigators with no underlying knowledge of the vehicle's perception system. To demonstrate the utility of this pipeline, we first analyze the latest information on AV crashes that have occurred in California and then select two crash scenarios that are analyzed in-depth using high-fidelity synthetic data generated from the automated vehicle simulator CARLA. The data visualization procedure is demonstrated on the real-world Kitti dataset by using the YOLO object detector and a monocular depth estimator called AdaBins. Depth from LIDAR is used as ground truth to calibrate and assess the effect of noise and errors in depth estimation. The visualization and data analysis from these scenarios clearly demonstrate the vast improvement in crash investigations that can be obtained from utilizing state-of-the-art sensing and perception systems used on AVs.

A Linear Objective Function-Based Heuristic for Robotic Exploration of Unknown Polygonal Environments.

This work presents a heuristic for describing the next best view location for an autonomous agent exploring an unknown environment. The approach considers each robot as a point mass with omnidirectional and unrestricted vision of the environment and line-of-sight communication operating in a polygonal environment which may contain holes. The number of robots in the team is always sufficient for full visual coverage of the space. The technique employed falls in the category of distributed visibility-based deployment algorithms which seek to segment the space based on each agent's field of view with the goal of deploying each agent into the environment to create a visually connected series of agents which fully observe the previously unknown region. The contributions made to this field are a technique for utilizing linear programming methods to determine the solution to the next best observation (NBO) problem as well as a method for calculating multiple NBO points simultaneously. Both contributions are incorporated into an algorithm and deployed in a simulated environment built with MATLAB for testing. The algorithm successfully deployed agents into polygons which may contain holes. The efficiency of the deployment method was compared with random deployment methods to establish a performance metric for the proposed tactic. It was shown that the heuristic presented in this work performs better the other tested strategies.

Modeling and numerical simulations of the influenced Sznajd model.

This paper investigates the effects of independent nonconformists or influencers on the behavioral dynamic of a population of agents interacting with each other based on the Sznajd model. The system is modeled on a complete graph using the master equation. The acquired equation has been numerically solved. Accuracy of the mathematical model and its corresponding assumptions have been validated by numerical simulations. Regions of initial magnetization have been found from where the system converges to one of two unique steady-state PDFs, depending on the distribution of influencers. The scaling property and entropy of the stationary system in presence of varying level of influence have been presented and discussed.