Apoorva Shende

Optimal Feedback Flow Rates for Pedestrian Evacuation in a Network of Corridors

This paper presents a methodology for the computation of optimal feedback flow rates (flow velocities and flow discharges) for pedestrian evacuation from a network of corridors using network-wide pedestrian congestion data. The pedestrian flow is defined in a macroscopic sense, wherein ordinary differential equations (ODEs) for each corridor and node are obtained using the conservation of pedestrian mass. The effect of congestion on the flow velocities and discharges in the corridor and the corridor intersections is explicitly modeled. Collectively, these corridor and node equations define the state-space model of the pedestrian flow in the network. The state variables signify the congestion in a corridor or an intersection, whereas the control variables directly affect the flow velocities and the flow discharges. For this model, an optimization-based control algorithm is developed to ensure a maximum total instantaneous input discharge that is subject to tracking the optimal congestion state and boundedness of the control variables. A comparison of the simulation results in the controlled and uncontrolled scenarios shows superior performance in the controlled case due to convergence to the optimal congestion state and consistently high network input and exit discharges.

A solution to the multiple aspect coverage problem

We introduce a novel task that arises in underwater search and inspection. In this task, an underwater vehicle must re-acquire and identify clusters of discrete objects. The challenge is to generate an efficient path for the vehicle given a probabilistic description of potential target locations. We propose an algorithm that generates an efficient path and show that the algorithm is superior to standard approaches.

Information Technology Requirements for Intelligent Evacuation Systems

In this paper we discuss the new concept of Intelligent Evacuation System (IES). This new technology will make a major change in evacuation strategies. Control, communication and computing technologies will be combined into IES system that can increase crowd safety and management without changing the physical structure of the facility. This paper outlines the key features of small and large area IES and shows how core crowd decisions are improved. We also propose a basic IES control system architecture and discuss the information technology issues related to its implementation.

Stochastic target interception in non-convex domain using MILP

In this paper we present a planning approach for the stochastic target interception problem, in which, a team of mobile sensor agents is tasked with intercepting multiple targets. We extend our previous work on stochastic target interception to non-convex domains and propose a cost that addresses minimum time requirement for probabilistically intercepting all the targets if possible over a finite horizon. Indeed, our optimization problem for the stochastic case has similar computational costs as the optimization program for the corresponding deterministic case. Our solution presumes that the system can be approximated by linear dynamics and Gaussian noise, with Gaussian localization uncertainty.

Optimal Control of Pedestrian Evacuation in a Corridor

The problem of optimal control of pedestrian evacuation from a corridor has been addressed. The corridor has been treated as a one dimensional link in the building network from which the pedestrians have to be evacuated. The governing flow equations are derived from the discretized continuity equation and a flow density relation for the pedestrian flow. Necessary conditions for the optimal control of these differential equations are developed using the calculus of variations method. The necessary conditions constitute a 2 point boundary value problem that has to be solved for the state, the co-state and the optimal control. Method of steepest descent is used to solve this problem. Numerical results are presented in the end for a test case.

Toward a mission value for subsea search with bottom-type variability

We propose a value function that can be used to evaluate candidate search paths for mobile sensor agents that seek to find specific objects. Our work is motivated by subsea applications where the mobile sensor agent surveys the seafloor using sonar. We presume that typical sensor performance characteristics are known, including probability of detection and probability of false alarm. Since variations in the environment can affect sensor performance, we also address the case that a stochastic description of the environment is available.

An approach to multi-agent area protection using bayes risk

We introduce a novel approach to controlling the motion of a team of agents so that they jointly minimize a cost function utilizing Bayes risk. We use a particle-based approach and approximations that allow us to express the optimization problem as a mixed-integer linear program. We illustrate this approach with an area protection problem in which a team of mobile agents must intercept mobile targets before the targets enter a specified area. Bayes risk is a useful measure of performance for applications where agents must perform a classification task. By minimizing Bayes risk, agents are able to explicitly account for the cost of incorrect classification. In our application, a team of mobile agents must classify potential mobile targets as threat or safe based on the likelihood the targets will enter the specified area. The agents must also maneuver to intercept targets that are classified as threat.

Multiple agent coordination for stochastic target interception using MILP

In this paper we present an approach to solving a stochastic multi-target interception problem. In the multi-target interception problem, a team of mobile sensors is tasked with intercepting a set of potential targets to reduce appropriately assigned damage cost. Our principal contribution is to express a stochastic version of the problem with a generalized cost as a mixed-integer linear program so that optimal sensor motion can be computed efficiently. Indeed, our optimization program for the stochastic problem has similar computational costs as the optimization program for the corresponding deterministic problem. Our solution presumes that the system can be approximated by linear dynamics and Gaussian noise, with Gaussian localization uncertainty.

Effects of correlation and seabed between multiple looks on the expected probability of detection of an underwater target

In this paper, we model an essential feature of underwater mine hunting with the aim of suggesting improvements to existing search tactics. We derive the effect of correlation between multiple sonar images of an underwater mine target - where a correlator represents the relation between images of the same target. For example, with duplicate images of the same target the correlator shows the absence of any additional information gain. However, when a contrasting image of a target is added to those already analyzed, the correlator proves the additional information gain. Our findings show that if correlation is weak, a specific set of equidistant angles will globally maximize the expected probability of detection.

Multi-agent motion planning using Bayes risk

We introduce a novel approach to controlling the motion of a team of agents so that they jointly minimize a cost function utilizing Bayes risk. Bayes risk is a useful measure of performance for applications where agents must perform a classification task, but is often difficult to compute analytically for many applications involving agent state variables. We use a particle-based approach that allows us to approximate Bayes risk and express the optimization problem as a mixed-integer linear program. By minimizing Bayes risk, agents are able to account explicitly for the costs associated with correct and incorrect classification. We illustrate our approach with a target interception problem in which a team of mobile agents must intercept mobile targets that are likely to enter a specified area in the near future. We show that the cooperative agent motion that minimizes a cost function utilizing Bayes risk is an efficient way to achieve selective interception. We introduce an approach to multi-agent motion control.Approach jointly minimizes a cost function utilizing Bayes risk for classification.We apply the framework in the context of target interception and collision avoidance.Framework uses a particle approach combined with mixed integer linear programming.