Autonomous Vehicle Trajectory Planning With Velocity-Based Steering Limits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating and optimizing trajectories for autonomous vehicles in complex driving environments with multiple interacting subsystems and dynamic changes is a challenging task due to the complexity of variables involved, including static and dynamic objects, weather, and traffic conditions.
Innovation Solution
The use of velocity-based steering limits, determined by a kinematic or dynamic model of the vehicle, which are converted into continuous functions for trajectory generation and optimization, allowing the vehicle to adjust its steering based on current state and environmental conditions, using cost functions to optimize navigation and safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional trajectory generation methods are used in complex driving environments, then the vehicle can navigate through static and dynamic objects, but the computational complexity and time required for trajectory optimization increase significantly
Solution Approach 1:
The patent segments the trajectory generation problem into multiple discrete steps: generating candidate trajectories, evaluating them against cost functions, and selecting the optimal one. This segmentation allows the system to manage computational complexity by breaking down the overall optimization task into smaller, more manageable sub-tasks that can be executed efficiently
Solution Approach 2:
The patent implements dynamic steering limits that adapt based on vehicle velocity and environmental conditions. Rather than using fixed steering constraints, the system dynamically adjusts steering angle limits and rate of steering change constraints according to the current driving context, allowing for more efficient computation while maintaining safety and comfort
2Reliability
If multiple subsystems and environmental variables are considered in trajectory generation, then navigation safety is improved, but the device complexity and computational load increase
Solution Approach 1:
The patent employs a universal cost function framework that can evaluate multiple different trajectory candidates using the same evaluation criteria. This multi-functional approach allows the system to consider various factors (safety, comfort, efficiency) through a unified evaluation mechanism, reducing the need for separate specialized subsystems for each consideration
Solution Approach 2:
The patent changes the parameter representation from fixed steering limits to velocity-based dynamic steering limits. By parameterizing the steering constraints as functions of vehicle velocity and environmental conditions, the system reduces complexity by using a smaller set of adaptive parameters rather than multiple fixed constraints across different operating conditions
3Ease of manufacture
If discrete steering limit data is used from lookup tables, then implementation is simplified, but the trajectory optimization becomes non-differentiable and computationally inefficient
Solution Approach 1:
The patent substitutes the discrete lookup table approach with a continuous mathematical function (neural network) that computes steering limits. This replacement transforms the system from a discrete, non-differentiable computation to a continuous, differentiable one, enabling the use of gradient-based optimization methods that are computationally more efficient while maintaining implementation feasibility
Data Source
AI summary
Techniques are described herein for generating trajectories for autonomous vehicles using velocity-based steering limits. A planning component of an autonomous vehicle can receive steering limits determined based on safety requirements and/or kinematic models of the vehicle. Discontinuous and discrete steering limit values may be converted into a continuous steering limit function for use during on-vehicle trajectory generation and/or optimization operations. When the vehicle is traversing a driving environment, the planning component may use steering limit functions to determine a set of situation-specific steering limits associated with the particular vehicle state and/or driving conditions. The planning component may execute loss functions, including steering angle and/or steering rate costs, to determine a vehicle trajectory based on the steering limits applicable to the current vehicle state.