Vehicle Trajectory Planning With Dynamic Candidate Action Diversity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face computational limitations in generating and considering all possible actions due to limited resources, leading to prohibitive latency and inadequate action diversity for safe navigation.
Innovation Solution
A planning system generates a set of candidate trajectories and actions that are dynamically varied laterally and temporally, using environmental and vehicle data to reduce the number of actions while maintaining diversity, thereby optimizing computational resources and reducing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the autonomous vehicle generates and considers every possible action within operational abilities, then the action diversity and safety are improved, but the computational latency becomes prohibitive
Solution Approach 1:
The planning system segments the action generation process into multiple stages: first generating a large pool of candidate actions, then filtering and ranking them through cost functions and tree search algorithms. This segmentation allows comprehensive action exploration while managing computational load through phased processing.
Solution Approach 2:
The system performs preliminary action generation and filtering before final decision-making. Candidate actions are pre-generated and pre-filtered through cost functions, so that when the actual control decision is needed, the computation is already substantially complete, reducing real-time latency.
2Productivity
If the autonomous vehicle reduces the number of possible actions computed, then the computational resources are optimized, but the action diversity may become insufficient for safe navigation
Solution Approach 1:
The system changes parameters dynamically by adjusting the number of candidate actions generated based on contextual factors such as vehicle speed, environment complexity, and safety requirements. This allows optimization of computational resources while maintaining sufficient action diversity for safe navigation in all operating conditions.
Solution Approach 2:
The action generation process is made dynamic rather than static. The system adaptively adjusts the breadth of action exploration based on real-time conditions, generating more actions when safety risks are higher and fewer actions when conditions are benign, thus balancing computational efficiency with safety requirements.
3Use of energy by moving object
If the autonomous vehicle uses a fixed set of candidate actions, then the computational resources are saved, but the system lacks adaptability to different operating conditions
Solution Approach 1:
The system employs dynamic adaptation where the candidate action generation process adjusts to different operating conditions. Rather than using a fixed set, the system modulates the number and type of actions generated based on vehicle speed, environmental factors, and safety requirements, achieving both efficiency and adaptability.
Data Source
AI summary
A vehicle with a planning system may generate a set of candidate trajectories for controlling the vehicle. The planning system may receive sensor data from sensors. The planning system may determine, based at least in part on the sensor data, environmental data, and/or the vehicle data. The planning system may determine, based at least in part on the environmental data and/or the vehicle data, a set of candidate trajectories corresponding to a set of contemporaneous candidate actions for the vehicle to follow. An individual candidate trajectory in the set is determined by dynamically shifting the individual candidate trajectory by a lateral distance relative to the vehicle or a lane in the environment. A candidate action may indicate a position, orientation, velocity, acceleration, or the like for the vehicle to track to by a certain time or distance.