Vision-Language Models (VLMs) have shown capabilities in interpreting visual content, but their reliability in safety-critical scenarios remains insufficiently explored. We introduce VERI, a diagnostic benchmark comprising 200 synthetic images (100 contrastive pairs) and an additional 50 real-world images (25 pairs) for validation. Each emergency scene is paired with a visually similar but safe counterpart through human verification. Using a two-stage evaluation protocol (risk identification and emergency response), we assess 17 VLMs across medical emergencies, accidents, and natural disasters. Our analysis reveals an "overreaction problem": models achieve high recall (70-100%) but suffer from low precision, misclassifying 31-96% of safe situations as dangerous. Seven safe scenarios were universally misclassified by all models. This "better-safe-than-sorry" bias stems from contextual overinterpretation (88-98% of errors). Both synthetic and real-world datasets confirm these systematic patterns, challenging VLM reliability in safety-critical applications. Addressing this requires enhanced contextual reasoning in ambiguous visual situations.

In the United States alone accidental home deaths exceed 128,000 per year. Our work aims to enable home robots who respond to emergency scenarios in the home, preventing injuries and deaths. We introduce a new dataset of household emergencies based in the ThreeDWorld simulator. Each scenario in our dataset begins with an instantaneous or periodic sound which may or may not be an emergency. The agent must navigate the multi-room home scene using prior observations, alongside audio signals and images from the simulator, to determine if there is an emergency or not. In addition to our new dataset, we present a modular approach for localizing and identifying potential home emergencies. Underpinning our approach is a novel probabilistic dynamic scene graph (P-DSG), where our key insight is that graph nodes corresponding to agents can be represented with a probabilistic edge. This edge, when refined using Bayesian inference, enables efficient and effective localization of agents in the scene. We also utilize multi-modal vision-language models (VLMs) as a component in our approach, determining object traits (e.g. flammability) and identifying emergencies. We present a demonstration of our method completing a real-world version of our task on a consumer robot, showing the transferability of both our task and our method. Our dataset will be released to the public upon this papers publication.

In this survey we are focusing on utilizing drone-based systems for the detection of individuals, particularly by identifying human screams and other distress signals. This study has significant relevance in post-disaster scenarios, including events such as earthquakes, hurricanes, military conflicts, wildfires, and more. These drones are capable of hovering over disaster-stricken areas that may be challenging for rescue teams to access directly. Unmanned aerial vehicles (UAVs), commonly referred to as drones, are frequently deployed for search-and-rescue missions during disaster situations. Typically, drones capture aerial images to assess structural damage and identify the extent of the disaster. They also employ thermal imaging technology to detect body heat signatures, which can help locate individuals. In some cases, larger drones are used to deliver essential supplies to people stranded in isolated disaster-stricken areas. In our discussions, we delve into the unique challenges associated with locating humans through aerial acoustics. The auditory system must distinguish between human cries and sounds that occur naturally, such as animal calls and wind. Additionally, it should be capable of recognizing distinct patterns related to signals like shouting, clapping, or other ways in which people attempt to signal rescue teams. To tackle this challenge, one solution involves harnessing artificial intelligence (AI) to analyze sound frequencies and identify common audio signatures. Deep learning-based networks, such as convolutional neural networks (CNNs), can be trained using these signatures to filter out noise generated by drone motors and other environmental factors. Furthermore, employing signal processing techniques like the direction of arrival (DOA) based on microphone array signals can enhance the precision of tracking the source of human noises.

Emergencies and critical incidents often unfold rapidly, necessitating a swift and effective response. In this research, we introduce a novel approach to identify and classify emergency situations from social media posts and direct emergency messages using an open source Large Language Model, LLAMA2. The goal is to harness the power of natural language processing and machine learning to assist public safety telecommunicators and huge crowds during countrywide emergencies. Our research focuses on developing a language model that can understand users describe their situation in the 911 call, enabling LLAMA2 to analyze the content and offer relevant instructions to the telecommunicator, while also creating workflows to notify government agencies with the caller's information when necessary. Another benefit this language model provides is its ability to assist people during a significant emergency incident when the 911 system is overwhelmed, by assisting the users with simple instructions and informing authorities with their location and emergency information.

Memes can sway people's opinions over social media as they combine visual and textual information in an easy-to-consume manner. Since memes instantly turn viral, it becomes crucial to infer their intent and potentially associated harmfulness to take timely measures as needed. A common problem associated with meme comprehension lies in detecting the entities referenced and characterizing the role of each of these entities. Here, we aim to understand whether the meme glorifies, vilifies, or victimizes each entity it refers to. To this end, we address the task of role identification of entities in harmful memes, i.e., detecting who is the 'hero', the 'villain', and the 'victim' in the meme, if any. We utilize HVVMemes - a memes dataset on US Politics and Covid-19 memes, released recently as part of the CONSTRAINT@ACL-2022 shared-task. It contains memes, entities referenced, and their associated roles: hero, villain, victim, and other. We further design VECTOR (Visual-semantic role dEteCToR), a robust multi-modal framework for the task, which integrates entity-based contextual information in the multi-modal representation and compare it to several standard unimodal (text-only or image-only) or multi-modal (image+text) models. Our experimental results show that our proposed model achieves an improvement of 4% over the best baseline and 1% over the best competing stand-alone submission from the shared-task. Besides divulging an extensive experimental setup with comparative analyses, we finally highlight the challenges encountered in addressing the complex task of semantic role labeling within memes.

The robotics industry had its fair share of memorable moments in 2022. Here we take a look back at our picks for the most memorable robotics stories of the year. The list is comprised of moments that made us laugh and cringe, as well as moments that left us surprised or amazed at the capabilities of various robots. Please let us know in the comments what you'll remember most from 2022. Subscribe to The Robot Report Newsletter to stay updated on the robotics stories you need to know about.

Collective behavior of people in large groups and emergent crowd dynamics can have dangerous and disastrous results when panic is introduced. These events can be caused by emergency situations such as fires in a large building or a stampeding effect when people are rushing in a densely packed area. In this paper, we will use an agent-based modeling approach to simulate different evacuation events in an attempt to understand what is the most efficient scenario. Specifically, we will focus on how people with disabilities are impacted by chosen parameters during an emergency evacuation. We chose an ABM to simulate this because we want to specify specific roles for different "agents" in our model. Specifically, we will focus on the influence of people with disabilities on crowd dynamics and the optimal exits. Does the placement of seating for people with disabilities affect the time it takes for the last person to exit the building? What effect does poor signage have on the time it takes for able-bodied and people with disabilities to exit safely? What happens if some people do not know about alternative exits in their panicked state? Using our agent-based model, we will investigate these questions while also adjusting other outside effects such as the density of the crowd, the speed at which people exit, and the location of people at the start of the simulation.

San Francisco leaders voted to allow the city's police department to use potentially lethal robots in emergency situations. "Under this policy, SFPD is authorized to use these robots to carry out deadly force in extremely limited situations when risk to loss of life to members of the public or officers is imminent and outweighs any other force option available," City Supervisor Rafael Mandelman wrote on Twitter. San Francisco leaders voted 8-3 on Tuesday in support of the new policy. The San Francisco Police Department has 17 robots, but none are armed with guns, and the department has no plans to do so. The department could deploy robots equipped with explosive charges "to contact, incapacitate, or disorient violent, armed, or dangerous suspect" during emergency situations when lives are at risk, according to a police department spokesperson.

Uncontrolled emissions of gases from industrial accidents and disasters result in huge loss of life and property. Such extreme events require a quick and reliable survey of the site for effective rescue strategy planning. To achieve these goals, a network of unmanned aerial vehicles can be deployed that survey the affected region and identify safe and danger zones. Although single UAV-based systems for gas sensing applications are well-studied in literature, research on the deployment of a UAV network for such applications, which is more robust and fault tolerant, is still in infancy. The objective of this project is to design a system that can be deployed in emergency situations to provide a quick survey and identification of safe and dangerous zones in a given region that contains a toxic plume without making any assumptions about plume location. We focus on an end-to-end solution and formulate a two-phase strategy that can not only guarantee detection/acquisition of plume but also its characterization with high spatial resolution. To guarantee coverage of the region with a certain spatial resolution, we set up a vehicle routing problem. To overcome the limitations imposed by limited range of sensors and drone resources, we estimate the concentration map by using Gaussian kernel extrapolation. Finally, we evaluate the suggested framework in simulations. Our results suggest that this two-phase strategy not only gives better error performance but is also more efficient in terms of mission time. Moreover, the comparison between 2-phase random search and 2-phase uniform coverage suggest that the latter is better for single drone systems whereas for multiple drones the former gives reasonable performance at low computational cost.

Scientists have created a robotic fabric that stiffens and relaxes in response to changes in temperature, which could be used in emergency situations. The material, developed at Yale University in the US, is equipped with a system of heat sensors and threads that stiffen to change the fabric's shape. Under heat changes, it can bend and twist to transform itself into adaptable clothing, shape-changing machinery and self-erecting shelters. Video footage shows the material going from a flat, ordinary fabric to a load-bearing structure supporting a weight, a model airplane with flexible wings and a wearable robotic tourniquet that activates in response to damage. 'We believe this technology can be leveraged to create self-deploying tents, robotic parachutes, and assistive clothing,' said Professor Rebecca Kramer-Bottiglio at Yale University.