Trajectory Prediction Using Posture Cues for Sudden Motion Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory prediction methods for pedestrians and vehicles primarily focus on historical motion data, neglecting the integration of time-series posture information and environmental context, leading to inaccuracies in predicting future trajectories.
Innovation Solution
A method that determines motion intention using time-series location and posture information, integrating environmental data through neural networks to enhance prediction accuracy by fusing these elements and iteratively refining future trajectories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only historical trajectory location information is used for prediction, then the prediction method is simple, but the prediction accuracy is insufficient especially in complex scenarios with sudden behavior changes
Solution Approach 1:
The prediction system segments the input data into multiple independent components: trajectory location information, posture information (including orientation of multiple parts), and environmental information. Each component is processed separately and then integrated, allowing the system to handle complex scenarios while maintaining manageable computational complexity
Solution Approach 2:
The system transitions from two-dimensional location coordinates to three-dimensional spatial information by incorporating posture data (orientation angles of multiple body parts) and environmental context. This dimensional expansion enables the model to capture sudden behavior changes that cannot be detected by location alone
2Measurement precision
If more input data including posture and environmental information is used, then prediction accuracy improves, but data processing complexity increases
Solution Approach 1:
The system merges multiple information sources (trajectory locations, posture data from multiple body parts, environmental context) into a unified prediction framework. By combining these diverse data types through a neural network model, the system achieves accurate motion intention recognition while managing the complexity through integrated processing
3Adaptability or versatility
If traditional historical trajectory methods are used, then computational resources are saved, but the ability to handle sudden behavior changes is insufficient
Solution Approach 1:
The system performs preliminary analysis by processing posture information and environmental data in parallel with trajectory data. This preliminary action enables the model to detect potential sudden behavior changes early, allowing for more adaptive predictions while optimizing computational resource usage through pre-computed features
Data Source
AI summary
Provided are a trajectory prediction method, an electronic device, and a storage medium. The method including that: a motion intention of an object is determined according to time-series location information and time-series posture information of the object, where the time-series location information is position information of the object at different time points within a preset time period, and the time-series posture information is posture information of the object at different time points within the preset time period, where the posture information at different time points includes orientation information of multiple parts of the object at the different time points; and a future trajectory of the object is determined according to the time-series location information, the time-series posture information and the motion intention.