Autonomous Vehicle Object Trajectory Prediction via Sensor-Map Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately predicting the motion of objects within their surroundings, which affects their navigation and safety, as existing systems struggle to integrate sensor data with geographic context effectively.
Innovation Solution
A computer-implemented method that fuses state data of objects with geographic area data to create a combined dataset, which is then input into a machine-learned model to predict object trajectories, enabling improved motion planning and control of autonomous vehicles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data is processed using traditional methods, then processing speed is maintained, but prediction accuracy of object trajectories deteriorates
Solution Approach 1:
The patent introduces geographic context data as an intermediary element that mediates between raw sensor data and trajectory predictions. This additional data layer helps the system understand environmental constraints (roads, sidewalks, obstacles) that influence object motion, thereby improving prediction accuracy without requiring fundamental changes to the core processing architecture
Solution Approach 2:
The patent adds a spatial dimension to the data processing by incorporating geographic information and map data alongside temporal sensor data. This multi-dimensional approach allows the system to predict trajectories by considering both historical motion patterns and environmental spatial constraints, improving accuracy while distributing computational complexity across different data dimensions
2Measurement precision
If geographic context data is integrated into the analysis, then prediction accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent segments the data processing into distinct modules: sensor data acquisition, geographic context data acquisition, data fusion, and trajectory prediction. Each module handles specific data types and processing tasks independently, which manages complexity by breaking down the integrated system into manageable, specialized components that can be developed and maintained separately
3Measurement precision
If machine-learned models are used for trajectory prediction, then prediction precision improves, but computational requirements increase
Solution Approach 1:
The patent performs preliminary processing of sensor data and geographic context data before inputting them to the machine-learned model. This includes filtering, feature extraction, and data fusion in advance, which reduces the complexity and size of data requiring intensive computational processing by the model, thereby lowering energy consumption while maintaining prediction precision
Data Source
AI summary
Systems and methods for predicting object motion and controlling autonomous vehicles are provided. In one example embodiment, a computer implemented method includes obtaining state data indicative of at least a current or a past state of an object that is within a surrounding environment of an autonomous vehicle. The method includes obtaining data associated with a geographic area in which the object is located. The method includes generating a combined data set associated with the object based at least in part on a fusion of the state data and the data associated with the geographic area in which the object is located. The method includes obtaining data indicative of a machine-learned model. The method includes inputting the combined data set into the machine-learned model. The method includes receiving an output from the machine-learned model. The output can be indicative of a predicted trajectory of the object.