Drenth, Rick (2026) Distributed Orientation-Based Formation Control with Regression-Based Vision Pose Estimation. Research Project, Industrial Engineering and Management.
|
Text
MasterResearchProjectRickDrenth08032026.pdf Download (5MB) | Preview |
Abstract
This thesis investigates orientation-aware distributed formation control under heading-constrained kinematics and its integration with monocular vision-based pose estimation. The main contribution is a corrective steering mechanism for gradient-based relative orientation control in systems with coupled translational and rotational dynamics. Classical gradient methods ensure convergence under decoupled assumptions, but heading constraints introduce undesired invariant sets where orientation errors vanish while distance errors persist. To address this, a corrective steering term is proposed that restores simultaneous regulation of both error types under rigidity assumptions. A Lyapunov-based analysis demonstrates local exponential convergence near the desired formation. The removal of invariant equilibria is analyzed and validated through deterministic simulations and Monte Carlo studies, showing consistent convergence and improved robustness to initial conditions. As a secondary contribution, a regression-based monocular pose estimation method for rectangular objects is developed. Pose recovery is formulated as nonlinear reprojection error minimization without explicit 2D–3D correspondences. An observability analysis highlights the effects of geometric symmetry, while simulations reveal anisotropic depth sensitivity. The estimator is positioned as part of a broader perception–control framework.
| Item Type: | Thesis (Research Project) |
|---|---|
| Supervisor name: | Jayawardhana, B. and Cao, M. and Rosa, M.R.R. |
| Degree programme: | Industrial Engineering and Management |
| Thesis type: | Research Project |
| Language: | English |
| Date Deposited: | 09 Apr 2026 14:05 |
| Last Modified: | 09 Apr 2026 14:05 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/37282 |
Actions (login required)
 |
View Item |