Autonomous Vehicle Feedforward Control Strategy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion planning algorithms for autonomous vehicles are either computationally intensive or not designed for multiple scenarios, making them inefficient for urban and highway driving.
Innovation Solution
The implementation of a lateral controller system for determining vehicle curvature and a longitudinal controller system for determining desired acceleration, using a control loop with a velocity error and feedforward term, to generate steering and throttle/braking commands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing motion planning algorithms are used for autonomous vehicles, then trajectory planning can be performed, but the processing is computationally intensive and inefficient
Solution Approach 1:
The motion planning system is divided into multiple independent modules: a trajectory generation module that creates reference trajectories, a trajectory optimization module that refines trajectories, and a feedforward control module that generates control commands. This segmentation allows each module to process specific aspects of motion planning independently, reducing overall computational load while maintaining processing efficiency.
Solution Approach 2:
The system pre-generates feedforward control commands based on the reference trajectory before actual execution. By calculating anticipated control actions in advance and storing them for quick retrieval, the system avoids performing complex computations in real-time, thereby improving processing efficiency while reducing instantaneous computational power requirements.
2Adaptability or versatility
If existing motion planning algorithms are used for autonomous vehicles, then trajectory planning can be performed, but the system is not designed for multiple different possible scenarios
Solution Approach 1:
The trajectory generation module creates universal reference trajectories that can serve as baseline paths for multiple different driving scenarios including urban streets, highways, and curved roads. The feedforward control module applies the same control architecture across all scenarios, adapting only the trajectory parameters rather than the fundamental algorithm structure, thereby achieving scenario versatility without proportionally increasing algorithm complexity.
Solution Approach 2:
The system applies scenario-specific optimizations at local levels rather than redesigning the entire algorithm for each scenario. The trajectory optimization module adjusts local trajectory parameters based on scenario characteristics while maintaining the overall algorithmic framework, allowing the system to handle multiple scenarios with controlled increases in complexity.
3Measurement precision
If a control loop with velocity error and feedforward term is used, then vehicle acceleration can be determined accurately, but the control system becomes more complex
Solution Approach 1:
The control loop continuously monitors velocity error between the actual vehicle state and the reference trajectory, using this feedback to adjust the feedforward acceleration commands. This feedback mechanism enables accurate acceleration determination by compensating for deviations in real-time, achieving high measurement precision while maintaining manageable control system complexity through the use of standard control loop architecture.
Data Source
AI summary
Systems and methods are provided for controlling an autonomous vehicle. A method includes using a lateral controller system for determining a vehicle's curvature. A longitudinal controller system is used for determining desired vehicle acceleration. The longitudinal controller system uses a control loop with respect to a velocity error and a feedforward term. Commands are generated based on the output of the lateral controller system and the longitudinal controller system.