Road User Trajectory Prediction With Type-Specific Feature Embedding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the future trajectories of road users in autonomous driving systems fail to consider the specific characteristics of different types of road users, leading to reduced prediction accuracy and reliability.
Innovation Solution
A method that transforms specific characteristics associated with predefined classes of road users, such as pedestrians and vehicles, into a fixed number of data elements using embedding algorithms, and allocates these to dynamic grid maps for enhanced trajectory prediction, incorporating unique features like body orientation for pedestrians and turning signals for vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the same features or characteristics are used to describe each road user regardless of type, then the data representation is simplified and uniform, but the prediction quality is reduced due to loss of specific characteristics
Solution Approach 1:
The patent segments the feature representation by creating separate embedding layers for different road user types (pedestrian, vehicle, cyclist). Each embedding layer processes type-specific characteristics independently, allowing the system to capture nuanced features like body orientation for pedestrians or turning signals for vehicles while maintaining a unified prediction framework.
Solution Approach 2:
The patent applies local quality by assigning different feature channels to different road user types within the same grid map. Specifically, pedestrians have channels for body and head orientation, vehicles have channels for heading angles and turning signals, and cyclists have channels for handlebar orientation. This allows each type to utilize its most relevant characteristics without being constrained by a uniform feature set.
2Measurement precision
If type-specific characteristics are incorporated for different road users, then the prediction quality is improved, but the device complexity increases due to multiple embedding layers
Solution Approach 1:
The patent achieves universality by designing a multi-functional embedding architecture where a single set of embedding layers serves multiple road user types simultaneously. The same neural network framework processes features for pedestrians, vehicles, and cyclists, with each type utilizing its designated channels. This unified approach reduces overall system complexity compared to having separate prediction models for each road user type.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method is provided for predicting trajectories of a plurality of road users. For each road user, a set of characteristics detected by a perception system of a vehicle is determined, wherein the set of characteristics includes specific characteristics associated with a predefined class of road users. The set of characteristics is transformed to a set of input features for a prediction algorithm via a processing unit of the vehicle, wherein each set of input data comprises the same predefined number of data elements. At least one respective trajectory for each of the road users is determined by applying the prediction algorithm to the input data.