DATE:

Friday, January 29, 2021

TIME:

3:30 PM

LOCATION:

Virtual Zoom Conference

SPEAKER:

Dr. Zahra Nili, Mechanical Engineering, San Diego State University

ABSTRACT:

Surveillance and exploration of adversarial and uncertain environments is a tedious task. In spaces with limited environmental cues, random-like search appears to be an effective approach as it allows the agents to perform an online coverage while avoiding intruders’ attacks. Inducing chaos into the agent’s controller system makes its navigation unpredictable and accounts for better scanning coverage. Unlike other random-like motion planning strategies (e.g., random-walk) which produce nondeterministic paths, chaos-based search strategies enable generation of deterministic but unpredictable trajectories. This quality allows the designer to control the paths and at the same time prevents the intruders from predicting the agents’ actions. Chaotic motions can be generated using certain nonlinear dynamical systems; a minute change in the parameters or initial conditions (ICs) of these systems can result in utterly different trajectories. Considering wide variety of dynamical systems and their corresponding range of parameters and ICs, the agents with chaotic path planners will have infinite number of choices; each influencing their performance and the unpredictability of their paths differently. For some of these systems, additional chaos control techniques can significantly improve the performance. The dynamical system properties are not the only important consideration in chaotic planners, the algorithm should as well decide about the method of coordination between the agents while taking into account the environment conditions (size, level of clutter, …), the agents characteristics (e.g., maneuverability), and the sensing and communications constraints. To optimize the swarm performance and realize adaptable, scalable, and effective search, we need strategies that allow exploring the trade-off space of different configurations to select the best settings as well as methods to better control the chaotic trajectories. In this talk, Dr. Nili will discuss the concepts related to chaotic motion planning, the trade-off space analysis study, the multi-agent coordination strategies, and the established techniques to control chaos.

Bio: Dr. Zahra Nili is an assistant professor in Mechanical Engineering Department and director of Dynamic Systems and Intelligent Machines Lab at San Diego State University (SDSU). She holds a Ph.D. from University of Quebec (École de Technologie Supérieure). Prior to this position, she was an assistant professor in Mechanical Engineering Department at Wichita State University. Her research interests include autonomy, cooperative robotics, and robot perception and learning. One area of focus is multi-agent coordination and nonlinear motion planning for mobile robots and unmanned air vehicles, with applications in surveillance and exploration missions. She is also working to establish acoustic perception technologies for robots essential in realizing effective human-robot interactions across physical structures and in enabling the agents to navigate and find occluded targets in dark and cluttered environments. Dr. Nili’s research has received support from NASA and AFRL/RI information institute.

HOST:

Satchi Venkataraman

VIDEO: