Autonomous Vehicle Intersection Exit Prediction via Aggregated Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting the exit of an intersection through which a moving obstacle is likely to exit is challenging for autonomous vehicles due to multiple possible exits and influences from surrounding lanes and other obstacles, making accurate path planning difficult.
Innovation Solution
A method that identifies and aggregates features of all possible exits, using recurrent neural networks (RNN) to determine the obstacle state evolution and intersection exit encodings, and calculates exit probabilities based on these features to plan a trajectory that avoids collisions with moving obstacles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional path planning methods are used to predict moving obstacle exits, then the system complexity remains low, but the prediction accuracy deteriorates due to multiple possible exits and influencing factors
Solution Approach 1:
The patent segments the prediction task into distinct components: obstacle feature extraction, exit feature extraction, encoding layers, and probability calculation. This modular segmentation allows each component to be optimized independently while maintaining overall system manageability, resolving the contradiction between accuracy and complexity
Solution Approach 2:
The patent introduces encoding layers as intermediary components that transform raw obstacle and exit features into compressed representations. These encodings serve as mediators between the complex input features and the final prediction, enabling accurate predictions while managing computational complexity through dimensionality reduction
2Measurement precision
If comprehensive features of all exits are considered for accurate prediction, then the prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent extracts only the most relevant features from each exit and aggregates them selectively rather than processing all possible exit information equally. This extraction approach maintains prediction accuracy by focusing on critical features while reducing computational power requirements by eliminating redundant processing
Solution Approach 2:
The patent merges the features of all possible exits into an aggregated representation that captures essential information about the intersection geometry and constraints. This merging consolidates multiple data sources into a unified feature set, achieving accurate predictions without the computational overhead of processing each exit independently
3Reliability
If the system accounts for all influencing factors including surrounding lanes and other obstacles, then the prediction reliability improves, but the processing time increases
Solution Approach 1:
The patent performs preliminary encoding of obstacle and exit features before the actual prediction step. By pre-processing and compressing the input features into encoded representations, the system prepares the data in advance, which speeds up the final prediction process while maintaining reliability through comprehensive feature consideration
Solution Approach 2:
The patent transforms the prediction problem by changing parameters from raw coordinate-based features to encoded feature representations. This parameter transformation reduces the dimensionality and complexity of the input data, enabling faster processing while preserving the essential information needed for reliable predictions
Data Source
AI summary
A moving obstacle such as a vehicle within a proximity of an intersection and one or more exits of the intersection are identified. An obstacle state evolution of a spatial position of the moving obstacle over a period of time is determined. For each of the exits, an intersection exit encoding of the exit is determined based on intersection exit features of the exit. An aggregated exit encoding based on aggregating all of the intersection exit encodings for the exits is determined. For each of the exits, an exit probability of the exit that the moving obstacle likely exits the intersection through the exit is determined based on the obstacle state evolution and the aggregated exit encoding. Thereafter, a trajectory of the ADV is planned to control the ADV to avoid a collision with the moving obstacle based on the exit probabilities of the exits.