Self-Driving Obstacle Avoidance Using Feasible-Area Potential Fields
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing Artificial Potential Field method for self-driving obstacle avoidance is prone to getting stuck in local minimums, fails to consider kinematic and dynamic constraints of self-driving vehicles, and lacks adaptability in structured urban environments.
Innovation Solution
A method that determines a feasible area around obstacles, calculates an attractive force towards a waypoint within this area using a preset attractive potential field function, and determines a repulsive force from obstacles using a preset repulsive potential field function, ultimately generating a steering angle control command to achieve obstacle avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing Artificial Potential Field method is applied for self-driving obstacle avoidance, then real-time performance and safety performance are improved, but the system gets stuck in local minimum and cannot sustain waypoint for moving forward
Solution Approach 1:
The patent segments the potential field calculation into multiple components: attractive potential fields for different waypoints, repulsive potential fields for different obstacles, and constraint potential fields for road boundaries. This segmentation allows the system to evaluate multiple possible paths simultaneously and select the optimal one, avoiding local minimum traps while maintaining safety performance.
Solution Approach 2:
The patent dynamically adjusts the potential field parameters based on the vehicle's current state, obstacle positions, and road constraints. The attractive and repulsive forces are recalculated in real-time, allowing the waypoint to be sustained and adjusted dynamically rather than being fixed, thus resolving the local minimum issue while maintaining safety.
2Device complexity
If existing Artificial Potential Field method is applied for self-driving obstacle avoidance, then sparse environment information is needed, but kinematic constraints and dynamic constraints of self-driving vehicle are not considered
Solution Approach 1:
The patent creates a universal potential field framework that simultaneously handles multiple functions: obstacle avoidance, road boundary adherence, kinematic constraint satisfaction, and dynamic constraint consideration. By integrating all these constraints into the potential field calculations, the system maintains sparse environment information requirements while achieving comprehensive adaptability to vehicle constraints.
Solution Approach 2:
The patent changes the parameters of the potential field functions to incorporate vehicle-specific constraints. The attractive and repulsive potential field functions are modified to include terms that represent kinematic and dynamic constraints, allowing the same framework to adapt to different vehicle types and operating conditions without requiring detailed environmental information.
3Shape
If existing Artificial Potential Field method is applied for self-driving obstacle avoidance, then smooth planning path is achieved, but adaptability to structured urban environment and reliability do not satisfy safety requirements
Solution Approach 1:
The patent implements feedback mechanisms where the potential field calculations continuously monitor the vehicle's position relative to obstacles, road boundaries, and constraint zones. The attractive and repulsive forces are adjusted based on this feedback, ensuring that the smooth planning path maintains high reliability by dynamically responding to changing environmental conditions and vehicle states.
Solution Approach 2:
The patent performs preliminary calculations of multiple candidate waypoints and their associated potential fields before selecting the final path. This preliminary action allows the system to pre-evaluate potential risks and adjust the planning path proactively, maintaining smoothness while enhancing reliability in structured urban environments.
Data Source
AI summary
The present disclosure provides a method and system for obstacle-avoidance in a self-driving, a system, and a storage medium, comprises: obtaining information of the obstacles, the information including sets of coordinates of the obstacles; determining a feasible area according to the information; determining a waypoint according to the feasible area, and determining an attractive force of the waypoint according to the waypoint and a preset attractive potential field function; determining a repulsive force of the obstacles according to the coordinates of the obstacles on sides of the feasible area and a preset repulsive potential field function; and determining a total force according to the attractive force and the repulsive force, and a steering command issued accordingly. A defect of Artificial Potential Field deflecting into the local minimum, and problems of environmental constraints and the constraints of the vehicle itself in a structured urban environment are resolved.