Occupancy and Flow Prediction Model for Autonomous Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning methods for predicting the position and velocity of surrounding objects in autonomous driving systems face challenges in achieving high prediction accuracy, particularly in dynamic environments where future positions and velocities can change due to driver actions.
Innovation Solution
A machine learning method and apparatus that input position data into a machine learning model to generate occupancy and flow data, with a process of expanding the occupancy region of ground truth data to calculate a loss parameter, which is used to update the model, thereby improving prediction accuracy by considering both occupancy and flow data simultaneously.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning methods are used to predict future position and velocity of surrounding objects, then prediction capability is provided, but prediction accuracy is insufficient in dynamic environments
Solution Approach 1:
The prediction task is segmented into two separate prediction targets: occupancy probability (whether an object will be at a location) and velocity vector (how fast and in what direction the object will move). This segmentation allows the model to optimize for accuracy in each aspect independently while adapting to dynamic driver actions through the dual-output architecture.
Solution Approach 2:
The invention adds a dimensional transformation by converting position prediction into occupancy probability mapping across a spatial grid, and by introducing velocity vectors as a separate dimension. This dimensional change enables the model to capture complex motion patterns and adapt to unpredictable driver behaviors more effectively.
2Reliability
If occupancy region expansion is performed on ground truth data, then loss calculation becomes more robust, but computational complexity increases
Solution Approach 1:
The occupancy region expansion is performed as a preliminary processing step on the ground truth occupancy data before loss calculation. By pre-expanding the occupancy regions, the model creates a more tolerant target distribution that accounts for inherent prediction uncertainties, making the loss calculation more robust without requiring complex adaptive algorithms during training.
Data Source
AI summary
A machine learning method includes: inputting pieces of position data to a machine learning model configured to receive the pieces of position data indicating a position of an object at respective first points in time, and output occupancy data and flow data, the occupancy data indicating occupancy probability of the object at a second point in time, the flow data indicating a velocity vector of the object at the second point in time; generating second ground truth occupancy data by performing a process of expanding an occupancy region of the object on first ground truth occupancy data at the second point in time; calculating a loss parameter, based on the occupancy data, the second ground truth occupancy data, the flow data, and ground truth flow data at the second point in time; and updating the machine learning model, based on the loss parameter.