Autonomous Vehicle Motion Planning Under Traffic Occlusions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion planning algorithms for autonomous agents in occluded environments are either too conservative, leading to performance degradation, or fail to integrate information gathering with risk-aware collision avoidance, lacking generality across different traffic scenarios.
Innovation Solution
A computer-implemented method that assesses occlusions using a scenario tree, determining a common motion plan until a decision postponing time based on predicted occlusions, allowing for passive and active information gathering, and adapting the decision postponing time as occlusions resolve.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If reachability analysis is used to predict all possible worst-case scenarios, then safety is improved, but vehicle performance degrades due to overly conservative behavior
Solution Approach 1:
The system dynamically adjusts the level of conservatism in motion planning by calculating a decision postponing time that adapts to current occlusion conditions. Instead of always assuming worst-case scenarios, the system postpones decisions until occlusions are resolved or time expires, allowing flexible transition between conservative and aggressive planning modes based on real-time sensor data and occlusion map updates.
Solution Approach 2:
The system changes the parameter of decision timing by introducing a dynamically calculated decision postponing time threshold. This parameter determines when to switch from deferred decision-making to active selection, allowing the system to adjust its risk tolerance and information gathering duration based on environmental conditions, thereby balancing safety and performance.
2Loss of information
If information-theoretic objectives are optimized during motion planning, then information gathering capability is improved, but risk-aware collision avoidance capability deteriorates
Solution Approach 1:
The system performs preliminary risk assessment by calculating decision postponing time based on predicted occlusion resolution before executing information gathering maneuvers. This preliminary action ensures that information gathering activities are bounded by safety constraints, preventing the vehicle from engaging in overly aggressive information gathering that could compromise collision avoidance.
Solution Approach 2:
The system uses feedback from updated sensor data and occlusion map changes to continuously monitor whether information gathering is proceeding safely. If new sensor data resolves occlusions earlier than predicted or reveals new hazards, the system adjusts the decision postponing time and motion plan accordingly, maintaining the balance between information gathering and collision avoidance.
3Reliability
If decision postponing is used to wait for occlusion resolution, then safety is improved, but time consumption increases
Solution Approach 1:
The system changes the time parameter dynamically by calculating a decision postponing time threshold based on predicted occlusion resolution. Instead of using fixed waiting periods, the system adapts the postponement duration to match actual environmental conditions, postponing decisions only as long as necessary for occlusion resolution while having a hard timeout to prevent excessive delays.
Solution Approach 2:
The system implements periodic reassessment of occlusion conditions during the decision postponing period. Instead of passively waiting for the full postponement duration, the system periodically checks for updated sensor data that may have resolved occlusions earlier than predicted, allowing early termination of the postponement period when no longer necessary, thus reducing time consumption while maintaining safety.
Data Source
AI summary
A computer-implemented method for determining a motion plan for driving an autonomous agent in a traffic situation, the agent including a sensor for capturing sensor data in a sensor field-of-view, the sensor data being indicative of the traffic situation.