The vision of Exalabs is to understand the complex dynamics of our world (including engineered, natural, physical, biological, and social systems), and create mechanisms that enable us to influence the behavior of large multi-scale systems with minimal intervention.
Self-organized (or patterned) behavior is prevalent in both engineered and natural complex systems, such as in traffic jams on highways, cascading failures in power grids, and epileptic seizures in neural systems, to name a few. There is significant environmental, productivity, and societal value associated with bringing improvements to such systems. The mission of Exalabs is to:
The goal of this project is to determine an analytical relationship between the ability of an agent to influence the self-organizing behavior of a complex system, and its location in state space. The problem is approached by studying the local controllability of a reduced-order linearized system that is indicative of the self-organized behavior exhibited large-scale complex systems. This technique has been demonstrated numerically for a simple stop-and-go wave in traffic, while analytical relationships are being developed for the N-coupled oscillators problem.
This project addresses the problem by identifying appropriate regions of the state space within which the control efforts exerted by a small set of agents can influence the self-organized dynamics of the ensemble. The problem is considered in the context of the ability of a small set of connected vehicles to affect the self-organized traffic jam dynamics. The methodologies adopted in this approach make use of the kinematic wave theory of traffic flow and the notion of controllability to present the novel concepts of influential subspaces and the null and event horizons. Results indicate that there exists a strong spatial dependence that governs an agent's ability to influence the self-organized dynamics of large-scale multi-agent systems. Specifically, under certain simplifying assumptions, the results indicate that connected vehicles must be able to communicate over a span of a few kilometers in order to impact traffic jam dynamics.
Vehicles in region 1 are too far away to impact the traffic jam, and vehicles in region 3 are too close. A vehicle in region 2 is in its influential subspace, and its control actions can impact the jam evolution.
As adaptive cruise control (ACC) technologies improve and make way into the mainstream vehicle market, it becomes necessary to study the impact of the interaction between ACC and human-driven vehicles on traffic flow. This project studies the impact of adaptive cruise control (ACC) algorithms on the formation of self-organizing traffic jams in medium-to-high density highway traffic. Currently, models of human and ACC driving behaviors are used within a master equation framework to model the traffic jams at mesoscopic scale. Further, mean field theory is used to reduce the probabilistic traffic jam dynamics to deterministic vehicle cluster or aggregate dynamics. Current work suggests that ACC results in higher traffic flows at the cost of increased susceptibility to congestion.
Analytical results (red dashed line) show that jams only form at higher densities as ACC penetration increases. Results from the Monte Carlo simulations agree with the analytical results. Contours indicate number of simulations (out of 1000) that resulted in a vehicular cluster (self-organized traffic jam).
Complex interconnected systems, such as transit buses and intelligent vehicles, are increasingly being manufactured with a large number of interacting components or subsystems. These components may interact with each other in ways that render the use of traditional analysis methods, such as Fault Trees (FTs) and Reliability Block Diagrams (RBDs), intractable or inefficient. This project assesses alternative data-driven techniques, such as Hidden Markov Models (HMMs), which may be more adept at handling complex systems. The Hidden Markov Modeling approach has been used to generate a stochastic reliability model for transit buses, which, along with the concept of ergodicity, can be used to determine the existence of component or subsystem dependencies in a complex system.
The proliferation of low-cost sensors has resulted in significant interest in the effects of such sensors on system performance. The goal of this project is to study the effect of sensor specifications and noise models on the performance of systems, such as the mission endurance of unmanned aerial vehicles and tracking accuracy of ground vehicles. Current approaches tackle these problems in the framework of Monte Carlo simulations, but upcoming work seeks to generate analytical results to address these issues.
Simulations in photo-realistic virtual environments help test the role of sensors and sensing equipment specifications on UAV mission capabilities.