Driving Path Adjustment for Occluded Object Uncertainty
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-driving vehicles face challenges in tracking objects that become occluded, as they lose sight of these objects, leading to potential safety risks due to the inability to account for their future positions and trajectories.
Innovation Solution
A system and method for tracking objects through occluded regions by identifying occlusion areas, adding uncertainty to object tracks within these areas, and adjusting the vehicle's driving path based on this uncertainty, using sensors, a planner-map, and a tracker to manage occlusion-related uncertainties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vehicle uses sensors to detect objects, then object detection capability is improved, but the ability to track occluded objects deteriorates because sensors cannot detect objects behind occlusions
Solution Approach 1:
The system performs preliminary actions by predicting object trajectories and maintaining track hypotheses before occlusion occurs. The tracker extrapolates object positions forward in time and creates predictive tracks that persist through occluded regions, allowing the system to maintain awareness of objects before they become undetectable
Solution Approach 2:
The system introduces an intermediary predictive modeling layer between sensor detection and path planning. This intermediary track hypothesis system acts as a mediator that bridges the gap between intermittent sensor detections and continuous navigation requirements, maintaining object presence information through occlusions using predicted trajectories and uncertainty propagation
2Reliability
If the vehicle maintains tracks of occluded objects, then safety is improved, but computational complexity increases due to uncertainty management
Solution Approach 1:
The system changes parameters by representing object uncertainty through probabilistic distributions and confidence scores rather than deterministic positions. Track hypotheses include uncertainty metrics that evolve over time, allowing the system to manage computational complexity through parameterized uncertainty representations that can be efficiently updated and propagated
Solution Approach 2:
The track hypotheses are dynamic structures that are continuously created, updated, and pruned based on new sensor measurements and predictions. The system dynamically adjusts track confidence levels and uncertainty parameters over time, allowing computational resources to be focused on the most relevant track hypotheses while maintaining safety through continuous re-evaluation
3Reliability
If the vehicle adjusts the driving path based on occluded object uncertainty, then collision avoidance is improved, but navigation efficiency deteriorates due to path adjustments
Solution Approach 1:
The system applies partial action by adjusting the driving path only to the extent necessary to account for occluded object uncertainty, rather than making excessive conservative adjustments. Path modifications are proportional to the predicted uncertainty and risk level, allowing the vehicle to maintain near-optimal trajectories while still ensuring safety through targeted, minimal path adjustments
Data Source
AI summary
Exemplary embodiments include methods and systems to adjust a driving path in a planner-map, including: identifying a driving surface; determining a path on the driving surface; identifying one or more objects relative to the driving surface using one or more sensors; detecting one or more occlusion areas from sensor data from the one or more sensors; tracking one or more tracks of one or more objects within the one or more occlusion areas; adding uncertainty to the one or more tracks within the one or more occlusion areas; and adjusting the path on the driving surface based on the uncertainty of the one or more tracks within the one or more occlusion areas.