Obstacle Avoidance Path Generation for Non-Holonomic Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-holonomic moving objects, such as vehicles and robots, face challenges in accurately following paths generated by path planning algorithms due to their driving and turning constraints, leading to potential collisions with obstacles.
Innovation Solution
A method involving a polynomial-based path generation algorithm to create a reference path, followed by a closed-loop controller using the Stanley algorithm to generate control values, and a kinematic model to fit the path into a feasible one, combined with a rapidly exploring random tree (RRT)-based algorithm to determine a final collision-free trajectory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a path planning algorithm generates a reference path, then the moving object can navigate from initial location to target location, but the non-holonomic moving object cannot accurately follow the path due to driving and turning constraints
Solution Approach 1:
The path generation process is divided into multiple stages: first generating a reference path using polynomial-based algorithm, then generating a feasible path by applying kinematic model corrections to account for non-holonomic constraints. This segmentation allows each stage to address specific requirements separately, ensuring both navigation capability and path following accuracy.
Solution Approach 2:
The system performs preliminary path planning to generate a reference path before actual movement, then pre-calculates a feasible path that incorporates the non-holonomic constraints of the moving object. This preliminary action ensures that the path is adapted to the object's capabilities before execution, preventing collision and improving followability.
2Stability of the object's composition
If a polynomial-based path generation algorithm is used to create a reference path, then the path can be generated smoothly, but the path may not be feasible for non-holonomic moving objects with driving constraints
Solution Approach 1:
The system uses a kinematic model to simulate the movement of the non-holonomic object along the reference path, calculating lateral error values and feedbacking this information to correct the control values. This feedback mechanism ensures that the generated feasible path accounts for the actual movement characteristics and constraints of the object, making the path both smooth and feasible.
Solution Approach 2:
The system modifies the control parameters (acceleration and steering values) based on the lateral error calculated from the kinematic model. By dynamically adjusting these parameters, the feasible path is generated to accommodate the non-holonomic constraints while maintaining the smoothness characteristics of the original polynomial-based reference path.
3Measurement precision
If the control value is corrected based on lateral error value, then the path following accuracy is improved, but the computational complexity increases
Solution Approach 1:
The closed-loop controller automatically calculates lateral error values and adjusts control values without requiring external intervention or complex manual tuning. The system serves itself by continuously monitoring its own performance and making real-time corrections, improving path following precision while keeping the control architecture relatively simple and self-contained.
Data Source
AI summary
A method for generating an obstacle avoidance path, performed by at least one processor includes generating a reference path from an initial location to a target location for a moving object to move, providing the generated reference path to a controller for generating a control value of the moving object to generate a control value corresponding to the reference path, providing the generated control value to a kinematic model related to the moving object to extract a state for the moving object to follow the reference path, and generating a feasible path corresponding to the reference path based on the extracted state.