Autonomous Vehicle Fallback Trajectory Planning for Safe Stops
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles (AVs) face challenges when their planning systems fail, leading to undesirable stops in unsafe locations and inadequate prediction of other objects' behavior, increasing the risk of dangerous situations.
Innovation Solution
AVs generate and update planned and fallback trajectories, with fallback trajectories considering predicted object behavior, allowing safe stopping locations when planning system failures occur.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the AV relies solely on the planned trajectory from the planning system, then the system complexity is reduced, but the reliability deteriorates when the planning system fails
Solution Approach 1:
The system pre-generates multiple fallback trajectories in advance, storing them for immediate use when the planning system fails. This preliminary preparation ensures that safe trajectories are available without requiring complex real-time recalculation during emergencies, thus improving reliability while managing complexity.
Solution Approach 2:
The system prepares multiple candidate fallback trajectories that account for various failure scenarios and object behaviors in advance. These pre-prepared trajectories act as a cushion or safety net, ensuring that when the planning system fails, the AV immediately has safe trajectory options without experiencing a reliability gap.
2Power
If the AV does not predict object behavior, then the computational load is reduced, but the safety deteriorates due to inadequate prediction of other objects' movements
Solution Approach 1:
The system applies different levels of prediction to different objects based on their relevance to the fallback trajectories. Instead of uniformly predicting all objects, it focuses computational resources on objects that could impact the predetermined stopping locations, thus reducing overall computational load while maintaining safety for critical objects.
Solution Approach 2:
The system performs partial prediction by only analyzing object behaviors that are relevant to the fallback trajectory scenarios. Rather than comprehensively predicting all possible object movements, it focuses on predictions necessary for safety, reducing computational power consumption while maintaining adequate safety coverage.
3Loss of time
If the AV stops immediately when planning system fails, then the response time is reduced, but the safety worsens due to stopping in unsafe locations
Solution Approach 1:
The system pre-identifies safe stopping locations and generates corresponding fallback trajectories before any failure occurs. When the planning system fails, the AV can immediately execute these pre-prepared trajectories to reach safe locations, achieving fast response time without compromising safety, as the safety assessment was already done in advance.
Solution Approach 2:
The system prepares multiple candidate fallback trajectories that lead to different potential stopping locations, cushioning against the risk of selecting an unsafe location. This preliminary preparation ensures that when failure occurs, the AV has pre-vetted safe stopping options ready, eliminating the trade-off between response time and safety.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to method, comprising: periodically receiving updates to a planned trajectory (310) and updates to a fallback trajectory (314) from a planning system (130) of an autonomous vehicle AV (100), wherein the planned trajectory (310) is a trajectory to a planned location of the AV and the fallback trajectory (314) is a trajectory to a fallback stopping location (312) in an environment of the AV; causing the AV to operate according to the updates to the planned trajectory (310); and upon determining that a threshold amount of time has passed since receiving a last update to the planned trajectory (310) from the planning system (130) of the AV, autonomously modifying operation of the AV according to a last update of the fallback trajectory (314) received from the planning system (130) of the AV.