Occlusion-Aware Motion Planning With Information-Gain Trajectories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion planning techniques for autonomous vehicles face challenges in ensuring safe trajectories when the vehicle's field of view is limited, leading to uncertainties due to unobserved regions, resulting in overly cautious and potentially unrecoverable maneuvers.
Innovation Solution
A method that computes collision probability scores and information gain scores for candidate trajectories, allowing the selection of a planned trajectory that balances safety with improved information gain for future planning horizons, using occupancy grids and a trained convolutional neural network to predict visibility and occupancy probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If motion planning assumes worst-case scenarios in unobserved regions, then safety is improved, but mobility deteriorates due to overly cautious maneuvers
Solution Approach 1:
The patent changes the parameter representation from binary (occupied/unoccupied) to probabilistic (occupancy probability values). This allows the system to express uncertainty quantitatively, enabling trajectories that balance safety and mobility by selecting paths with acceptable risk levels rather than avoiding all unobserved regions entirely.
Solution Approach 2:
The patent introduces an intermediary computational layer (the processor executing the method) that mediates between the limited sensor data and trajectory selection. This intermediary computes occupancy probabilities and collision risks, transforming uncertain sensor inputs into informed trajectory decisions that neither overly conservative nor reckless.
2Reliability
If the vehicle follows a conservative trajectory to avoid unobserved regions, then collision risk is reduced, but information gain about the environment is lost
Solution Approach 1:
The patent implements feedback by using updated sensor observations to revise occupancy probability estimates dynamically. As the vehicle moves and new regions become observable, the system updates its belief state about unobserved regions, creating a closed-loop system that continuously improves environmental understanding while managing risk.
Solution Approach 2:
The patent makes the occupancy grid dynamic by continuously updating occupancy probability values as new sensor data arrives. This dynamic representation allows the system to adapt its risk assessment in real-time, enabling the vehicle to exploit newly gained information to adjust its trajectory and gain further information strategically.
3Device complexity
If traditional cost functions are used for trajectory evaluation, then computational simplicity is maintained, but accuracy in assessing collision risk deteriorates
Solution Approach 1:
The patent replaces traditional mechanical/rule-based cost function evaluation with a probabilistic computational model. Instead of using fixed rules and simple cost additions, the system uses probability theory to compute collision risks, substituting deterministic mechanics with statistical computation to achieve more accurate risk assessment.
Data Source
AI summary
Systems, methods and computer-readable media for selecting a trajectory for an autonomous vehicle are disclosed that include computing a current vehicle state for the autonomous vehicle based on observations by a sensing system; computing respective collision probability scores for a plurality of candidate trajectories based on the current vehicle state; computing respective information gain scores for the plurality of candidate trajectories based on the current vehicle state, the information gain score for each candidate trajectory indicating an respective information gain for a next planning horizon interval that is subsequent to the current planning horizon interval; and selecting a planned trajectory from the plurality of candidate trajectories based on the respective collision probability scores and respective information gain scores.