Path planning for first responders in the presence of moving obstacles

Navigation services have gained much importance for all kinds of human activities ranging from tourist navigation to support of rescue teams in disaster management. However, despite the considerable amount of route guidance research that has been performed, many issues that are related to navigation for first responders still need to be addressed.

During disasters, emergencies can result in different types of moving obstacles (e.g., fires, plumes, floods), which make some parts of the road network temporarily unavailable. After such incidents occur, responders have to go to different destinations to perform their tasks in the environment affected by the disaster. Therefore they need a path planner that is capable of dealing with such moving obstacles, as well as generating and coordinating their routes quickly and efficiently.

During the past decades, more and more hazard simulations, which can modify the models with incorporation of dynamic data from the field, have been developed. These hazard simulations use methods such as data assimilation, stochastic estimation, and adaptive measurement techniques, and are able to generate more reliable results of hazards. This would allow the hazard simulation models to provide valuable information regarding the state of road networks affected by hazards, which supports path planning for first responders among the moving obstacles.

The objective of this research is to develop an integrated navigation system for first responders in the presence of moving obstacles. Such system should be able to navigate one or more responders to one or multiple destinations avoiding the moving obstacles, using the predicted information of the moving obstacles generated from by hazard simulations. In this dissertation, the objective we have is expressed as the following research question:

How do we safely and efficiently navigate one or more first responders to one or more destinations avoiding moving obstacles?

To address the above research questions, this research has been conducted using the following outline: 1). literature review; 2). conceptual design and analysis; 3). implementation of the prototype; and 4). assessment of the prototype and adaption. We investigated previous research related to navigation in disasters, and designed an integrated navigation system architecture, assisting responders in spatial data storage, processing and analysis.Within this architecture, we employ hazard models to provide the predicted information about the obstacles, and select a geo-database to store the data needed for emergency navigation. Throughout the development of the prototype navigation system, we have proposed:

• a taxonomy of navigation among obstacles, which categorizes navigation cases on basis of type and multiplicity of first responders, destinations, and obstacles;
• a multi-agent system, which supports information collection from hazard simulations, spatio-temporal data processing and analysis, connection with a geo-database, and route generation in dynamic environments affected by disasters;
• data models, which structure the information required for finding paths among moving obstacles, capturing both static information, such as the type of the response team, the topology of the road network, and dynamic information, such as changing availabilities of roads during disasters, the uncertainty of the moving obstacles generated from hazard simulations, and the position of the vehicle;
• path planning algorithms, which generate routes for one or more responders in the presence of moving obstacles. Using the speed of vehicles, departure time, and the predicted information about the state of the road network, etc., three versions (I, II, and III) of Moving Obstacle Avoiding A* (MOAAStar) algorithms are developed: 1). MOAAstar— I/Non-waiting, which supports path planning in the case of forest fires; 2). MOAAstar—II/Waiting, which introduces waiting options to avoid moving obstacles like plumes; 3). MOAAstar—III/Uncertainty, which can handle the uncertainty in predictions of moving obstacles and incorporate the profile of responders into the routing.

We have applied the developed prototype navigation system to different navigation cases with moving obstacles. The main conclusions drawn from our applications are summarized as follows:

• In the proposed taxonomy, we have identified 16 navigation cases that could occur in disaster response and need to be investigated. In addressing these navigation problems, it would be quite useful to employ computer simulations and models, which can make reliable predicted information about responders, the targets, and obstacles, in finding safe routes for the responders.
• The approach we provide is general and not limited to the cases of plumes and fires. In our data model, the data about the movement of hazards is represented as moving polygons. This allows the data model to be easily adjusted to merge and organize information from models of different types of disasters. For example, the areas that are affected by floods can also be represented as moving polygons. To facilitate the route calculation, not only the data of obstacles but also the information about the state of road networks affected by obstacles need to be structured and stored in the database.
• In planning routes for responders, the routing algorithms should incorporate the dynamic data of obstacles to be able to avoid the hazards. Besides, other factors, such as the operation time of tasks, the required arrival time, and departure time, also need to be considered to achieve the objectives in a rescue process, e.g., to minimize the delays caused by the moving obstacles.
• The profile of responders is quite important for generation of feasible routes for a specific disaster situation. The responders may have different protective equipment that allows them to pass through different types of moving obstacles, and thus can have different classification of risk levels to define the state of the road network. By taking into account the profile of the responders, the navigation system can propose customized and safe routes to them, which would facilitate their disaster response processes.

On the basis of our findings, we suggest the following topics for future work:

• As presented Wang and Zlatanova (2013c), there are still a couple of navigation cases that need to be addressed, especially the ones that involve dynamic destinations. More algorithms would be needed to solve these navigation problems. Besides, some extreme cases (e.g., the obstacle covers the target point during the course of an incident) also need to be investigated.
• Using standard Web services, an Android navigation application, which can provide navigation services in the environment affected by hazards, needs to be developed and tested in both the daily practice and real disasters. In this application, a user interface with various styling options should also be designed for different situations, e.g., waiting and moving, day and night, and urgent and non-urgent.
• Because the communication infrastructure may not be available or work properly during a disaster response, a decentralized method is needed to allow different users to negotiate with each other and to make local agreements on the distribution of tasks in case there is no support from the central planning system. Another type of multi-agent system would be needed to handle this situation.
• Introduce variable traveling speed into the re-routing process. The vehicle speed plays an important role in generation of routes avoiding moving obstacle, and can be influenced by many factors, such as the obstacles, the type of vehicles, traffic conditions, and the type of roads. Therefore, it would be needed to investigate how to derive the current and future speed from trajectories of vehicles.
• Apply the system to aid navigation in various types of natural disasters, using different hazard simulation models (e.g., flood model). More types of agents would be needed and integrated into the system to handle heterogeneous data from these models. Extensions of the data model are also required to meet a wider range of informational needs when multiple disasters occur simultaneously.