Object Orientation Modes for Stable Vehicle Trajectory Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle trajectory planning systems face challenges in accurately determining the orientation of dynamic objects due to incomplete or ambiguous sensor data, leading to potential abrupt changes in vehicle operation and safety hazards.
Innovation Solution
A vehicle computing system uses multiple hypotheses parameter determination techniques to integrate sensor data from various modalities, determining parameter modes and associated probabilities to predict object trajectories, thereby reducing ambiguity and improving operational safety.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data from multiple modalities is integrated to determine object orientation, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the object orientation determination into distinct parameter modes (e.g., first mode, second mode, third mode) corresponding to different sensor modalities or data characteristics. Each mode is processed independently through the machine-learned model, allowing the system to handle multiple sensor types without creating a monolithic complex processing pipeline. This segmentation enables modular integration of sensor data while maintaining measurement precision.
2Reliability
If multiple parameter modes with probabilities are used to predict object trajectories, then reliability is improved, but computational complexity increases
Solution Approach 1:
The patent transforms the trajectory prediction problem by changing the parameter representation from single deterministic values to multiple parameter modes with associated probabilities. The machine-learned model processes these probabilistic parameter modes to generate multiple possible trajectories, each with a confidence level. This parameter transformation improves reliability by accounting for uncertainty while the probabilistic framework provides a systematic way to manage computational complexity through weighted processing of multiple hypotheses.
3Reliability
If abrupt trajectory changes are reduced through probabilistic modeling, then vehicle safety is improved, but response time may be reduced
Solution Approach 1:
The patent applies preliminary action by pre-training the machine-learned model on extensive trajectory data before deployment. During real-time operation, the model leverages its pre-learned patterns to quickly evaluate multiple parameter modes and predict trajectories without requiring extensive computational resources for training. This preliminary preparation enables the system to reduce abrupt trajectory changes through probabilistic modeling while maintaining acceptable response times for vehicle control decisions.
Data Source
AI summary
Systems and techniques for determining more accurate object parameter values, such as orientation values (e.g., yaw, location, position, etc.), for objects detected in an environment are disclosed. A vehicle computing system may receive sensor data from multiple sensor systems that indicates a parameter value determined at the individual sensor systems. The vehicle computing system may determine values and probabilities for particular parameters modes based on the sensor data parameter values and may further filter these values using a mixture model to determine probability distributions for the modes and associated values. These filtered values and modes may then be used to determine predicted object trajectories that can be used to control a vehicle.