Grant: Discovery Launch Supplement
École Polytechnique de Montréal
$12,500.00
A scalable framework for nonlinear observer design in robotic systems
FY 2024-25
Summary
Latent variables are prevalent in engineering and science, often caused by the inadequacy of measurements due to cost and physical constraints. Although they are not directly accessible, accurately determining latent variables is indispensable for robotic systems where one aims to obtain real-time information for decision-making, monitoring, and planning, or to design a controller that interacts with the real physical world. Achieving this requires high-performance estimation algorithms, commonly referred to as state observers, to reconstruct online reliable estimates of latent variables in an optimal sense.
The state-of-the-art design of nonlinear observers follows two main methods: analytical and optimization-based. The former uses extended dynamics to asymptotically construct unknown states but can be difficult for general nonlinear systems. On the other hand, optimization-based approaches reconstruct states at each instant by minimizing some cost function based on historical measurements. However, these methods require substantial time and computing resources for processing data and solving online large-scale optimization problems, which can significantly impact real-time implementation. In robotics, system states typically live on manifolds, and these geometric constraints pose special challenges such as nonlinearity, large-scale complexity, uncertainty, and weak observability due to non-informative trajectories.
This program aims to advance a new observer framework designed to enhance the online estimation of latent variables in nonlinear systems, making it more computationally scalable, accurate, and robust, with a special emphasis on robotics. Our approach distinguishes itself from the existing two categories of observers and will be built upon a new approach, parameter estimation-based observer (PEBO), developed by our team. The PEBO approach is still in its infancy but has shown great promise through a number of preliminary results in recent years. This research seeks to fully realize the potential and generalization of PEBO. The main objective is to propose promising estimation techniques that seamlessly integrate analytical and optimization-based observer approaches, ensuring both provable global stability and real-time implementation. The research will advance our understanding and fundamental knowledge of observer/observability theory and provide state-of-the-art algorithms for engineering practice. While the focus of the program is on robotic estimation, the theoretical foundation and general-purpose estimation algorithms developed can be applied to address complex challenges in various fields, ranging from mechatronics and power electronics to aerospace. The program will train 3 PhD, 2 MSc, and 5 undergraduate students with special expertise in control, estimation, and robotics, all of which are in high demand to support priority areas in Canada's technology industry.
Program
Discovery Launch Supplement
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
DGECR-2024-00400
Name nom
Yi, Bowen
Department département
Département de génie électrique
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
12500
Program id
DGECR
Program name en
Discovery Launch Supplement
Program name fr
Supplément Tremplin vers la découverte
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1510
Committee name en
Electrical and Computer Engineering
Committee name fr
Génie électrique et informatique
Area of application code
1206
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Engineering
Area of application fr
Génie
Research subject code
2509
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Control systems
Research subject fr
Asservissements
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
A scalable framework for nonlinear observer design in robotic systems
Keywords
state estimation of dynamical systems; nonlinear observers; nonlinear control systems; robotic estimation; control theory; mechatronics; robotics; navigation and localization; autonomous systems
Application summary
Latent variables are prevalent in engineering and science, often caused by the inadequacy of measurements due to cost and physical constraints. Although they are not directly accessible, accurately determining latent variables is indispensable for robotic systems where one aims to obtain real-time information for decision-making, monitoring, and planning, or to design a controller that interacts with the real physical world. Achieving this requires high-performance estimation algorithms, commonly referred to as state observers, to reconstruct online reliable estimates of latent variables in an optimal sense.
The state-of-the-art design of nonlinear observers follows two main methods: analytical and optimization-based. The former uses extended dynamics to asymptotically construct unknown states but can be difficult for general nonlinear systems. On the other hand, optimization-based approaches reconstruct states at each instant by minimizing some cost function based on historical measurements. However, these methods require substantial time and computing resources for processing data and solving online large-scale optimization problems, which can significantly impact real-time implementation. In robotics, system states typically live on manifolds, and these geometric constraints pose special challenges such as nonlinearity, large-scale complexity, uncertainty, and weak observability due to non-informative trajectories.
This program aims to advance a new observer framework designed to enhance the online estimation of latent variables in nonlinear systems, making it more computationally scalable, accurate, and robust, with a special emphasis on robotics. Our approach distinguishes itself from the existing two categories of observers and will be built upon a new approach, parameter estimation-based observer (PEBO), developed by our team. The PEBO approach is still in its infancy but has shown great promise through a number of preliminary results in recent years. This research seeks to fully realize the potential and generalization of PEBO. The main objective is to propose promising estimation techniques that seamlessly integrate analytical and optimization-based observer approaches, ensuring both provable global stability and real-time implementation. The research will advance our understanding and fundamental knowledge of observer/observability theory and provide state-of-the-art algorithms for engineering practice. While the focus of the program is on robotic estimation, the theoretical foundation and general-purpose estimation algorithms developed can be applied to address complex challenges in various fields, ranging from mechatronics and power electronics to aerospace. The program will train 3 PhD, 2 MSc, and 5 undergraduate students with special expertise in control, estimation, and robotics, all of which are in high demand to support priority areas in Canada's technology industry.