Generating and Optimizing Topologically Distinct Guesses for Mobile Manipulator Path Planning with Path Constraints
Rufus Cheuk Yin Wong, Mayank Sewlia, Adrian Wiltz, Dimos V. Dimarogonas
AI summary
Problem
Gradient-based optimal path planning for mobile manipulators frequently converges to suboptimal local minima due to nonconvex obstacle and end-effector path constraints. Existing methods lack reliable mechanisms to generate diverse initial guesses that span different topological path classes.
Approach
The authors reduce the high-dimensional configuration space to a low-dimensional graph and apply a modified Neighborhood Augmented Graph Search (NAGS) algorithm to discover homotopically distinct paths. These topologically diverse paths are then used as initial guesses for nonlinear trajectory optimization to identify and select the best multi-local optimum.
Key results
- Dimensionality-reduced configuration graph tailored for end-effector path constraints
- Modified NAGS algorithm that accurately detects homotopy classes despite tiny obstacles and non-uniform discretization
- Pipeline that consistently generates multiple homotopically distinct initial guesses for trajectory optimization
- Simulation validation showing improved multi-local optimality and robustness over baseline planners
Why it matters
Provides a practical framework for industrial mobile manipulators to reliably solve complex constrained path planning tasks without getting trapped in local minima.
Abstract
Optimal path planning is prone to convergence to local, rather than global, optima. This is often the case for mobile manipulators due to nonconvexities induced by obstacles, robot kinematics and constraints. This paper focuses on planning under end effector path constraints and attempts to circumvent the issue of converging to a local optimum. We propose a pipeline that first discovers multiple homotopically distinct paths, and then optimizes them to obtain multiple distinct local optima. The best out of these distinct local optima is likely to be close to the global optimum. We demonstrate the effectiveness of our pipeline in the optimal path planning of mobile manipulators in the presence of path and obstacle constraints.