Mobile Robot Trajectory Planning With Guided Constrained Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Constrained optimization-based planners for mobile robots face challenges in achieving fast convergence to globally optimal trajectories while adhering to hard constraints, particularly due to local optima issues and slow convergence times, especially as the complexity of cost functions and models increases.
Innovation Solution
A two-stage optimization approach using a trained function approximator to provide initialization data for a runtime optimizer, where the first stage uses a simpler cost function and dynamics model to guide the second stage towards a globally optimal solution, ensuring compliance with hard constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-linear constrained optimization solvers are used to solve the planning problem, then the solution can satisfy hard constraints, but the convergence is slow and uncertain and may converge to local optima
Solution Approach 1:
The patent applies preliminary action by using a trained function approximator to generate an initial trajectory before the optimization process begins. This pre-computed initialization guides the non-linear constrained optimization solver to converge faster to a globally optimal solution rather than starting from arbitrary initial conditions, thereby reducing convergence time while maintaining constraint satisfaction.
Solution Approach 2:
The function approximator serves as an intermediary between the scenario description and the optimization solver. It translates scenario parameters into informed initial trajectory estimates, bridging the gap between problem formulation and optimization execution, which accelerates convergence and reduces the risk of getting trapped in local optima.
2Manufacturing precision
If the complexity of cost function and mobile robot model increases, then the quality of trajectory improves, but the convergence becomes slower and more uncertain
Solution Approach 1:
The function approximator performs preliminary computation to generate informed initial trajectories that account for complex cost functions and robot models. By pre-processing the complex problem structure, the optimization solver can focus on fine-tuning rather than exploring from scratch, maintaining high trajectory quality while reducing computational burden during real-time execution.
3Measurement precision
If non-linear constrained optimization solvers are used, then globally optimal solutions can be found, but the solvers tend to converge to local optima
Solution Approach 1:
The trained function approximator performs preliminary action by generating high-quality initial trajectory estimates that are close to the global optimum. This initialization significantly reduces the search space for the non-linear constrained optimization solver, making it much less likely to get trapped in local optima and substantially improving the reliability of finding globally optimal solutions.
Data Source
Figure 1~2B
Figure 3
Figure 4
AI summary
A computer-implemented method of planning a mobile robot trajectory, the method comprising: receiving a set of scenario description parameters describing a scenario and a desired goal for the mobile robot therein; computing, from the set of scenario description parameters, initialization data defining an initial mobile robot trajectory; initializing a runtime optimizer with the initialization data; computing, by the runtime optimizer, a final mobile robot trajectory that substantially optimizes a cost function for the scenario, subject to a set of hard constraints that the final mobile robot trajectory is guaranteed to satisfy, the initial mobile robot trajectory being an approximation of the final mobile robot trajectory, wherein initializing the runtime optimizer guides the optimizer from the initial mobile robot trajectory to the final mobile robot trajectory that satisfies the hard constraints.