1. General information
This component of the research program shall be undertaken by a PhD candidate to develop advanced algorithms and calibration methods so that commercial grade MEMS inertial sensors can be used to acquire precise attitude for an Army soldier-navigation-solution (SNS).
2. Research Project Proposal
Principle objective of this research program is to develop advanced attitude estimation algorithms and inertial MEMS sensor calibration methods to establish attitude estimation limits on low cost commercial grade MEMS inertial sensor triad assemblies. To achieve these objectives, a PhD student shall gain expertise over a period of three years in several disciplines namely: nonlinear estimation, Bayesian calibration methodologies for creation of high accuracy/fidelity inertial sensor error models, application of nonlinear geometric control theory for attitude stabilization and precision performance. Some benefits outlined here of using MEMS inertial sensors for this research project are: 1) a smaller footprint, 2) lower weight and power consumption, 3) higher volume of production at a lower cost, 4) lower part count resulting in lower system maintenance. These are attractive attributes for product development particularly for a soldier-navigation-solution (SNS) whose need is expressed by SSTRM research program. Research emphasis shall be placed on the development of estimation algorithms solely dependent on inertial sensors without the need for external aiding sources which typically increases costs, system size, weight and power requirements.
Using inertial and magnetic measurements: angular rates from gyros, incremental velocity changes from accelerometers and earth’s magnetic field vector respectively, an algorithm shall be formulated by combining a nonlinear attitude estimator perhaps using concepts from fuzzy logic, implemented as a quaternion entity for computational efficiency. An adaptive nonlinear control loop shall be designed to exploit the dynamical environment of such sensors whose gains can be tuned in real-time as the dynamical conditions experienced by these sensors changes via the soldiers trajectory. Standard tests shall be designed to assess the attitude performance of these algorithms and compared with baseline performance metrics of a conventional Kalman filter approach. Algorithm performance shall be demonstrated with and without external aiding sources to justify technical feasibility and technology commercialization for the intended product.
To satisfy schedule and cost constraints, this PhD student shall also have an opportunity to develop expertise in the use of Matlab/Simulink toolset considered a de-facto standard worldwide for modern engineering product analysis and synthesis using Model Based Design (MBD) methodologies, for the purposes of rapid prototyping, requirements analysis, software verification and validation per conventional commercial and military standards. Computational loading analysis of these models should be quantified so that cost benefit analysis can be performed for the selection of hardware platforms using the above tools.
Some outcomes of this research program could be but not limited to: 1) to develop an algorithms that can provide reasonable attitude estimates without additional aiding when required, 2) develop high fidelity sensor error inertial models using Bayesian calibration methodologies, 3) design a generic nonlinear sensor adaptive control system that can be reused seamlessly for other plant models, 4) expansion of Matlab/Simulink proprietary toolset capabilities currently available at ETS laboratories.
A suggested chronological research roadmap is provided subject to minor alterations as required. First the doctoral student needs to explore previous work (literature review) in the area of attitude estimation, then develop a detailed calibration procedure for the purposes of creating high fidelity sensor error models, to remove/calibrate in-house sensor biases, scale factor, broadband noise, misalignments and other unobservable modeling errors whilst using conventional and Bayesian modeling methods. Initial sensor parameters can be established by using manufacturer’s specification data sheets or alternatively measuring sensor outputs and conducting Allan variance analysis techniques to quantify the type of noise these sensors exhibit. Followed by, based on preprocessing of sensor measurements an algorithm such as a complementary nonlinear filter shall be developed to estimate the noise of these pre-calibrated inertial sensors in the special orthogonal group, finally a nonlinear attitude control system shall be designed to keep this system stable under various dynamical conditions. This research program shall conclude with an extensive report documenting and demonstrating test results with simulated and real world data sets, these results shall be compared with currently available standard Kalman filter algorithms for attitude estimation to demonstrate enhanced precision and robustness of low cost MEMS attitude sensing systems.
To transform this research effort into a commercially viable proposition for technology transfer, a cost benefit analysis and market needs analysis for military and commercial sectors should be provided for low cost attitude sensing systems, to justify the initial investment of costs by the government and industry partners. A suggested product diversification portfolio plan should also be provided so that multiple product lines can be created from this common core technology developed by our PhD student under the supervision of ETS faculty.
In conclusion this research opportunity at ETS provides our doctoral student to investigate and learn theoretical intricacies involved with the unobservable and uncontrollable aspects of this dynamical sensor inertial IMU model, it exposes the student with modern scientific tools that are available at the students disposal and how they can be successfully used to solve this innovative and an important research problem as expressed by SSTRM by using advanced modern differential geometric nonlinear control theory successfully applied to solve problems of such low cost physical systems.
3. Importance for the partner
Capstone report has listed new technologies which are important for the Canadian Soldier. Numerica Technologies is working on a larger project of ETS actual project contribution. The determination of the position of a target for a soldier is of great importance for our partner. The soldier should be equipped with light devices, efficient technologies, precise and robust embedded systems. To achieve this main goal, precise and robust attitude determination algorithm should be developed. Such attitude determination system should have all these qualifications so that the soldier will use it with minimal complexity. For example only, the end result device may be integrated within the helmet of the soldier. An optical monocular lenses could be deployed with a cross target in the center of the optical such that the soldier can fix his desired object/target. The electronic within the system should be capable of computing in real-time the precise attitude and position of the target such that this information could be relayed to other team element part of the soldier infrastructure. ETS will play a portion of this role to focus its research on attitude determination (Project Phase I). The ranging determination and analysis will be cover by another research group of Numerica.
4. Interaction with industry
The student will work in collaboration with his professor, professional research engineer, the rest of the team and our partner collaborator. He will spend 50% of his time in the industry partner Numerica whose office is based in Montreal. The main industrial project supervisor will be Mr. Pierre Gosselin
5. Deliverables and Schedule
Year #1:
First version of literature review report (DGA-1031 : Report #1)
Year #2 :
Final research proposal (DGA-1033: Report #2)
Analysis and simulation results (Report #3)
Year #3 :
Comparison of algorithms and selection of best candidate (Report #4)
Implementation on prototype with performance analysis (Report #5)
PhD thesis
Note: First year of PhD student will be focused on PhD examination for (DGA-1031, DGA-1032 and DGA-1033)
