Vehicle Obstacle Avoidance via Linear Constraint Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing advanced driver assistance systems (ADAS) face challenges in efficiently minimizing the likelihood of contact between a vehicle and obstacles by actuating brake and steering systems, particularly in complex scenarios requiring rapid decision-making.
Innovation Solution
The system uses a computer to receive a nominal input specifying acceleration and steering angle, and then determines a final input modification by minimizing a cost function subject to constraints represented as linear inequalities in a two-dimensional space, focusing on obstacles around the vehicle.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If standard quadratic programming solvers are used to minimize vehicle-obstacle contact likelihood, then the optimization accuracy is improved, but the processing time increases
Solution Approach 1:
The patent segments the continuous optimization problem into discrete evaluation points along constraint boundaries. Instead of solving the full quadratic programming problem continuously, the system evaluates the cost function at specific discrete points (vertices and boundary points) of the feasible region defined by linear constraints, thereby reducing computational complexity while maintaining adequate optimization accuracy
Solution Approach 2:
The patent uses simplified linear constraints as approximations of the actual nonlinear safety boundaries. These linear constraints act as computationally inexpensive surrogate models that can be quickly evaluated, providing a fast though approximate solution to the optimization problem without requiring heavy computational resources
2Measurement precision
If the system evaluates multiple potential input modifications to minimize the cost function, then the optimization quality is improved, but the computational complexity increases
Solution Approach 1:
The patent divides the feasible solution space into discrete evaluation points along the constraint boundaries. By segmenting the continuous optimization problem into specific vertices and boundary points, the system can evaluate a manageable number of candidate solutions without needing to explore the entire continuous space, thus reducing computational complexity while maintaining optimization quality
Solution Approach 2:
The patent evaluates the cost function at specific critical points (vertices and boundary points) rather than exhaustively searching all possible solutions. This partial evaluation approach focuses computational effort on the most promising regions of the solution space, achieving adequate optimization quality without the excessive computational burden of complete enumeration
Data Source
AI summary
A computer includes a processor and a memory, and the memory stores instructions executable by the processor to formulate a plurality of constraints, determine a final input modification to a nominal input that minimizes a cost function subject to the constraints, and actuate a component of a vehicle according to the nominal input and the final input modification. Each constraint indicates a respective obstacle relative to the vehicle. Each constraint is represented as a linear inequality in a two-dimensional space with dimensions for acceleration and steering angle. The nominal input includes a nominal acceleration and a nominal steering angle. The final input modification includes a final change to the nominal acceleration and a final change to the nominal steering angle. The computer determines the final input modification by individually calculating the cost function for a plurality of potential input modifications that are on the constraints.