Research

The goal of this topic is to develop advanced guidance, navigation, and control algorithms for unmanned vehicles (UxV). The developed algorithms enables UxV to autonomously perform the specified goals while avoiding obstacles discovered in their paths. These activities are a part of a larger effort to establish a theoretical foundation for autonomous and cooperative multi-UxV guidance solutions in adversarial environments. Recently, the focus is to explore multi-vehicle collaboration scenarios such as (i) sharing map information data between platforms lacking obstacle detection sensors and (ii) construction of environment map using multiple vehicles. It is planned to investigate possible solutions of achieving real-world implementation of developed avoidance algorithms. 

This topic focuses on developing novel methods for early detection of faults to provide avoidance of system failures, breakdowns, and catastrophes. Especially for safety-critical systems such as nuclear reactors, chemical plants, aerospace vehicles, autonomous vehicles, unmanned vehicles or fast rail systems, fault detection and isolation are of great importance. FDI algorithms are used to identify any occurrence of a fault in the system (detection) and to pinpoint the type and location of the fault (isolation). While there are various types of FDI development methods, our research has been focused on redundant-sensor-based and data-driven algorithms.

The focus of this topic is moving object detection and tracking, SLAM, 3D reconstruction using computer vision techniques. We develop novel algorithms for several applications such as, intruder aircraft detection using stereo cameras, 3D reconstruction of environment by swarm of UAVs using VisualSfM, motion detection and tracking using 360-degree camera. We also focus on visual SLAM algorithms for mobile robot navigation.

This topic includes evolutionary based novel algorithm and bio-inspired computational method development. One of the algorithms developed is called the trait-based heterogeneous populations plus (TbHP+) which is a variant of the genetic algorithm (GA) model. The developed TbHP+ model employs a memory concept in the form of immunity and instinct to provide the populations with a more efficient guidance. Another focus is implementation of lateral inhibition (LI), which is the most important mechanism to occur in biological distributed sensory networks. LI relies on one simple principle; each sensor strives to suppress its neighbors in proportion to its own excitation. LI mechanism is utilized in different applications to localize the unknown position of a source in a sensory network system.

The aim of this topic is to design and evaluate a distributed behavior methodology and test scalability for high-throughput heterogeneous multi-agent teams. Our focus is to develop theoretical foundation of integrated distributed behavior and role & task assignment architecture. We attempt to correlate entropy measure with the scalability, quality, and reliability of the collaborative multi-agent teams. A type of generalized entropy, i.e. Tsallis entropy, is used for two main purposes (i) distributed behavior modeling (ii) dynamic role assignment & conflict resolution.