Articulated Object Motion Prediction With a Joined State Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous vehicle planning systems face inefficiencies in predicting the future states of articulated objects, such as trucks with trailers, as they require processing multiple possible positions and velocities for each object separately, leading to increased computational resources and reduced accuracy.
Innovation Solution
Implementing a 'joined motion model' that leverages the mathematical relationship between the portions of an articulated object, using filtering algorithms like Kalman filters to predict the states of both the tractor and trailer based on the state data of the tractor, thereby reducing computational resources and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple possible positions and velocities are processed separately for each object, then prediction accuracy is maintained, but computational resources increase
Solution Approach 1:
The patent combines the prediction of tractor and trailer states into a unified filtering process. Instead of separately processing multiple possible positions and velocities for each object, the system integrates them into a single joint state prediction using Kalman filtering, thereby reducing computational resources while maintaining prediction accuracy through the mathematical relationship between connected portions.
Solution Approach 2:
The filtering algorithm serves multiple functions simultaneously: it predicts the state of the tractor, predicts the state of the trailer, and enforces the mathematical relationship between them. This multi-functionality eliminates the need for separate processing of each object's possible positions and velocities, reducing computational overhead while maintaining comprehensive prediction accuracy.
2Measurement precision
If separate processing is used for each object portion, then individual object accuracy is maintained, but overall system complexity increases
Solution Approach 1:
The patent merges the prediction systems for the tractor and trailer into a single integrated filtering process. The unified approach maintains individual object prediction accuracy by incorporating the mathematical relationship between portions, while simultaneously reducing system complexity by eliminating redundant separate processing mechanisms.
3Ease of operation
If traditional filtering is applied without considering mathematical relationships, then processing is simpler, but prediction accuracy for articulated objects decreases
Solution Approach 1:
The patent modifies the filtering process by incorporating the mathematical relationship between articulated object portions as an additional constraint parameter. This changes the filtering equations to include the kinematic relationship between tractor and trailer, thereby improving prediction accuracy for articulated objects while maintaining the fundamental simplicity of the filtering approach.
Data Source
AI summary
A vehicle computing system may implement techniques to determine whether two objects in an environment are related as an articulated object. The techniques may include applying heuristics and algorithms to object representations (e.g., bounding boxes) to determine whether two objects are related as a single object with two portions that articulate relative to each other. The techniques may include predicting future states of the articulated object in the environment. One or more model(s) may be used to determine presence of the articulated object and/or predict motion of the articulated object in the future. Based on the presence and/or motion of the articulated object, the vehicle computing system may control operation of the vehicle.