+1 438 506 4945
PUBLISHED WORK
2018 - Ongoing
2018
EXPERIMENTAL VERIFICATION OF UAS BASED BATTERY TERMINAL VOLTAGE COLLAPSE DETECTION ON A SIMPLE EMBEDDED PLATFORM.
This paper experimentally verifies that the Universal Adaptive Stabilization (UAS) based Li-ion battery terminal voltage collapse strategy can be implemented on a simple embedded platform i.e. microcontrollers. Due to the complex nature of Nussbaum type switching functions, it is usually assumed that using the UAS strategy with certain Nussbaum functions is very computationally intensive and requires extremely powerful computation equipment due to the small step-size requirements. However, in this paper we present the implementation and verification of a universal adaptive stabilizer, which uses a MittagLeffler function as a Nussbaum function, and implement it on a simple microcontroller based system, i.e. the PIC24FJ128GA010 which is a focal point in this paper. The adaptive observer has proven to be capable of tracking the terminal voltage of different batteries in several separate trials. In another set of experiments, it is shown that bypassing the low-pass filter may also provide alternate ways to detect Li-ion battery terminal voltage collapse.
2019
TECHNIQUES OF INDOOR POSITIONING SYSTEMS (IPS): A SURVEY
Indoor Localization has been a challenging task in robotics for years, due to the obstruction of the GPS signals. Although multiple techniques have been proposed, determining the best among them is still highly dependent on applications and no best solution is agreed on. One technique, researchers and developers are adopting –especially in the robotics field-, is the visual/inertial sensor fusion. However, the high number of these systems and their diversity, makes selecting the adequate one for an application a challenging task. This paper aims at presenting an exhaustive list indoor positioning techniques with more emphasis on visual/depth sensors, along with some recommendations for selected applications.
2021
ALLEVIATING DYNAMIC MODEL UNCERTAINTY EFFECTS FOR IMPROVED BATTERY SOC ESTIMATION OF EVS IN HIGHLY DYNAMIC ENVIRONMENTS
Dynamic battery modeling uncertainties, even if low, may lead to significant performance degradation or even divergence of the state-of-charge estimation algorithm. This paper investigates the integration of the extended-Kalman-filter with the smooth-variable-structure-filter algorithms for state-of-charge estimation of lithium-ion batteries. The robustness of the presented approach to modeling uncertainty is assessed for batteries operating in highly dynamic environments. The presented approach combines the benefit of the smooth- variable-structure-filter in its robustness to model uncertainty with the benefit of the extended-Kalman-filter in its near-optimality for a given dynamics and measurement noise sequences. The algorithm is rigorously tested using various datasets including standardrized and artificial drive cycles with added dynamics. The drive cycle power profile is calculated for an electric Ford F150 truck and scaled for the 18650PF cell used in the tests. Experimental validation is performed by investigating four different scenarios in which knowledge of the initial conditions as well as accuracy of the battery model were varied. The results demonstrate a substantially enhanced estimation accuracy achieved by the adopted approach through its optimality to measurement and model noise as well as its robustness to model uncertainty. The adopted approach results in a reduction in the complexity of the state-of-charge control algorithm and therefore enhances the battery management system.
2022
DEPTH/INS SENSOR FUSION USING CONSTRUCTIVE NEURAL NETWORKS
Submitted