Research

Engineering Sciences

Title :

Development of advanced non-linear non-Gaussian state estimators with application to real-life problems

Area of research :

Engineering Sciences

Principal Investigator :

Dr. Rahul Radhakrishnan, National Institute Of Technology Calicut (NITC), Kerala

Timeline Start Year :

2022

Timeline End Year :

2025

Contact info :

Equipments :

Details

Executive Summary :

The proposal aims to address a long-standing research problem in state estimation, which involves determining the unknown state of a system from noise-corrupted system measurables. The conventional method is to develop a state estimation approach in the Bayesian framework, but this approach is not suitable for non-linear systems due to the presence of fluctuations and spikes in real-life sensor measurements. State estimators designed to accommodate Gaussian noises may fail to generate accurate estimates when fed with measurements with non-Gaussian noises. In target tracking problems, measurements generated by radar or sonar are corrupted with glint noise, shot noise, or a combination of both, which must be modeled as pure non-Gaussian random variables. This calls for the development of new state estimators that can effectively tackle non-Gaussian noises in sensor measurements to provide an accurate estimate of the unknown state of the system. The proposed research will focus on reformulating the conventional state estimation approach to effectively handle non-Gaussian noises, relaxing the Gaussian assumption and MMSE criterion. The main motive is to find, propose, and reformulate the conventional estimation framework, finding an alternative to MMSE that captures more information from a non-Gaussian density. The newly developed advanced estimation algorithms will be rigorously tested and validated through simulation experiments for target tracking, continuous glucose and insulin estimation, and state of charge estimation. Prototype or hardware implementation will be carried out for testing the estimation accuracy of the proposed algorithms with real/synthetic measurement data. The outcome of this proposal will positively impact target tracking communities, health sciences, and battery management systems development for electric vehicles (EVs), as it provides a solution for handling non-Gaussian measurement noises.

Co-PI:

Dr. Shambhu Nath Sharma, Sardar Vallabhbhai National Institute Of Technology, Surat, Gujarat-395007, Dr. Gangireddy Sushnigdha, Sardar Vallabhbhai National Institute Of Technology, Surat, Gujarat-395007

Total Budget (INR):

21,29,718

Organizations involved