Ego Vehicle Behavior Planning With Predicted Sensor Visual Range
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep learning-based prediction and planning approaches for autonomous vehicles ignore the limitations of sensor visibility, leading to increasing uncertainty in occupancy predictions due to unseen areas, which can be unsafe.
Innovation Solution
Integrate a visual range model into the prediction process using a machine learning model, such as a convolutional neural network, to account for changes in sensor visibility influenced by the traffic scene, and incorporate attention mechanisms to focus on relevant areas, enhancing the prediction and planning of driving maneuvers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning-based prediction approaches use occupancy maps as intermediate representations, then planning trajectories can be computed, but the limited visual range of sensors is ignored leading to increasing uncertainty of predicted occupancy over time
Solution Approach 1:
The patent segments the prediction task into two distinct components: occupancy prediction and visual range prediction. By dividing the overall prediction system into these separate modules, each can be optimized independently - occupancy prediction focuses on object location while visual range prediction specifically models sensor visibility constraints, thereby resolving the uncertainty issue without compromising overall accuracy
Solution Approach 2:
The patent introduces visual range prediction as an intermediary component that mediates between the occupancy map and the final planning decision. This intermediary explicitly models the sensor's limited visual range and uses it to adjust or mask occupancy predictions in obscured areas, thereby maintaining reliability over time by accounting for visibility constraints
2Ease of operation
If the sensor system operates with limited visual range due to occlusions, then the area that can be viewed is restricted, but deep learning approaches do not account for this limitation causing unsafe planning
Solution Approach 1:
The patent applies preliminary action by predicting visual range constraints before they affect occupancy predictions. The system proactively models future visual range limitations based on predicted traffic scene developments, and uses this information to adjust occupancy predictions in advance, ensuring safe planning decisions are made before occlusions cause uncertainty
Solution Approach 2:
The patent implements a feedback mechanism where visual range prediction results are fed back into the occupancy prediction process. The predicted visual range information continuously adjusts the occupancy predictions by masking or reducing confidence in obscured areas, creating a closed-loop system that maintains safety by constantly accounting for sensor visibility constraints
Data Source
AI summary
A computer-implemented method is for planning the behavior of an ego vehicle as part of a traffic scene. The ego vehicle is equipped with an in-vehicle sensor system whose visual range is influenced by a respective current traffic scene. The method includes generating a scene representation of the current traffic scene using scene-specific sensor data captured by the in-vehicle sensor system, and predicting a future development of the traffic scene based on the generated scene representation. The method further includes planning driving maneuvers taking into account the prediction of the future development of the traffic scene. The method predicts the effects of the future development of the traffic scene on the visual range of the in-vehicle sensor system.