Occluded Region Occupancy Reasoning for Smoother Autonomous Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous vehicle navigation techniques often result in overly cautious behavior, leading to frequent braking and discomfort for passengers due to the inability to accurately assess occluded regions in the environment, which can be either static or dynamic, causing uncertainty about the vehicle's actions.
Innovation Solution
An occlusion reasoning system that uses sensor data and map information to determine the likelihood of occluded regions being unoccupied by dynamic objects, classifying them as 'pseudo-visible' based on observation over time, allowing the vehicle to make more informed decisions and traverse the environment more efficiently and safely.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional autonomous vehicle navigation techniques are used, then safety is maintained through conservative decision-making, but passenger comfort deteriorates due to frequent unnecessary braking
Solution Approach 1:
The patent introduces an occlusion reasoning system as an intermediary between sensor data and navigation decisions. This system uses map information and observation time to infer the occupancy state of occluded regions, providing a bridge that allows the vehicle to make more informed decisions without requiring direct sensor observation of all areas, thereby reducing unnecessary conservative braking while maintaining safety
Solution Approach 2:
The system performs preliminary reasoning about occluded regions before the vehicle reaches them. By using map information and observing visible regions over time to predict occupancy states in advance, the vehicle can plan smoother trajectories and avoid last-minute conservative braking maneuvers, improving passenger comfort while maintaining safety
2Measurement precision
If conventional sensor-based detection is used, then direct observation of visible regions is achieved, but uncertainty about occluded regions causes overly conservative behavior
Solution Approach 1:
Instead of directly detecting occluded regions through sensors, the patent inverts the approach by inferring occupancy states of occluded regions from observations of visible regions and map information. The system reasons backward from what is visible to conclude what is likely hidden, transforming a detection limitation into a reasoning opportunity that reduces information loss about occluded areas
Solution Approach 2:
The system uses feedback from observing visible regions over time to update beliefs about occluded region occupancy. By continuously monitoring entry and exit points of occluded regions and using this feedback with map information, the system refines its occupancy state estimates, reducing uncertainty without requiring direct sensor observation of the occluded areas themselves
3Reliability
If the vehicle waits for complete visibility of occluded regions before acting, then safety is improved, but productivity decreases due to delayed decision-making
Solution Approach 1:
The patent applies partial action by making navigation decisions based on partial information about occluded regions rather than waiting for complete visibility. The occlusion reasoning system provides probabilistic occupancy states that enable the vehicle to proceed with cautious confidence, achieving sufficient safety assurance without the excessive delay of waiting for full observation, thus improving traversal efficiency
Solution Approach 2:
The system performs preliminary assessment of occluded region occupancy using map information and visible region observations before the vehicle reaches those areas. This advance reasoning allows the vehicle to plan and execute maneuvers with appropriate confidence levels, avoiding both premature action and excessive waiting, thereby improving productivity while maintaining safety
Data Source
AI summary
Techniques are discussed for controlling a vehicle, such as an autonomous vehicle, based on occluded areas in an environment. An occluded area can represent areas where sensors of the vehicle are unable to sense portions of the environment due to obstruction by another object or sensor limitation. An occluded region for an object is determined, along with one or more visible regions proximate the occluded region. Entry and/or exit regions may be determined based on known directions of traffic and/or drivable surface boundaries. Based on a threshold speed, the vehicle can designate portions of the occluded region as pseudo-visible. Additionally, if a dynamic object traverses through the occluded region from the entry region, a portion of the occluded region may be considered pseudo-visible. Pseudo-visibility may also be determined based on movement of an occluded area. The vehicle can be controlled to traverse the environment based on the pseudo-visibility.