Autonomous Vehicle Controller Model Switching for Reverse Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving dynamic models lack sufficient control accuracy for low-speed, tight-turning scenarios, particularly when driving forward or reverse, due to differences in vehicle dynamics that are not adequately accounted for.
Innovation Solution
A method-switching approach is employed, where a forward driving model (e.g., bicycle model) and a reverse driving model (hybrid dynamic and kinematic model) are selected based on gear position, combined with predictive feedback models (look-ahead and look-back models) and augmented with Fourier Transform analysis to improve control efforts for lateral and heading errors, using a Linear Quadratic Regulator (LQR) for precise control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single forward dynamic model (bicycle model) is used for all driving conditions, then the model is simple and easy to implement, but control accuracy deteriorates in low-speed reverse driving scenarios
Solution Approach 1:
The patent implements a dynamic model switching mechanism that adapts the vehicle dynamics model based on driving conditions (forward/reverse gear position and speed thresholds). When reverse gear is detected and speed is below a threshold, the system switches from a standard bicycle model to a specialized reverse driving model that accounts for reversed wheel rotation and altered friction characteristics, thereby maintaining control accuracy across different operating conditions without requiring a permanently complex model structure
Solution Approach 2:
The patent changes key dynamic parameters (friction coefficients, steering geometry effects, center of mass positioning) based on driving direction and speed conditions. In low-speed reverse scenarios, the model adjusts parameters to reflect reversed tire friction characteristics and altered steering dynamics, allowing the same controller structure to achieve accurate control by adapting parameters rather than requiring fundamentally different control algorithms
2Ease of operation
If maximum turning angle is used in tight turning scenarios, then the vehicle can complete tight turns and U-turns, but wheel-friction force deteriorates affecting control precision
Solution Approach 1:
The patent implements a predictive feedback control mechanism that continuously monitors lateral and heading errors and adjusts steering commands in real-time. The look-ahead predictive model anticipates future position deviations based on current tracking errors, allowing the system to compensate for friction losses at maximum steering angles by proactively adjusting control inputs before significant deviations occur, thereby maintaining control precision even during aggressive tight turning maneuvers
3Device complexity
If conventional motion planning estimates path difficulty only from curvature and speed, then the planning is simple, but accuracy deteriorates because vehicle type differences are not considered
Solution Approach 1:
The patent applies local quality by tailoring the motion planning and control parameters to specific vehicle characteristics and driving directions. Rather than using a universal planning approach, the system adjusts path estimation and control parameters based on whether the vehicle is in forward or reverse gear, and adapts to vehicle-specific properties such as steering axis geometry and center of mass location, thereby improving path estimation accuracy for different vehicle types and operating conditions
Data Source
AI summary
Generating control effort to control an autonomous driving vehicle (ADV) includes determining a direction (forward or reverse) in which the ADV is driving and selecting a driving model and a predictive model based upon the direction. In a forward direction, the driving model is a dynamic model, such as a “bicycle model,” and the predictive model is a look-ahead model. In a reverse direction, the driving model is a hybrid dynamic and kinematic model and the predictive model is a look-back model. Current and predicted lateral error and heading error are determined using the driving model and predictive model, respectively. A linear quadratic regulator (LQR) uses the current and predicted lateral error and heading errors to determine a first control effort An augmented control logic determines a second, additional, control effort, to determine a final control effort that is output to a control module to drive the ADV.