Trajectory Association Using Machine-Learned Models for Occlusion Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately associating and tracking the trajectories of objects over time, particularly due to occlusions and environmental constraints, which can lead to errors in motion planning and safety.
Innovation Solution
A computer-implemented method using a machine-learned model, such as LSTM, to generate temporally-consistent trajectories by associating and refining object trajectories across different time steps, incorporating association scores and life cycle management (birthing, coasting, reaping) to account for occluded objects and ensure accurate motion planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional trajectory tracking methods are used, then computational resources are consumed, but trajectory tracking errors increase due to occlusions and environmental constraints
Solution Approach 1:
The system performs preliminary actions by predicting future trajectories of objects before actual occlusions occur. The machine-learned model generates predicted trajectories for multiple time steps ahead, allowing the system to prepare for potential occlusions and maintain accurate tracking by having pre-computed trajectory information ready when occlusions happen.
Solution Approach 2:
The patent introduces an intermediary machine-learned model that acts as a mediator between raw sensor data and trajectory tracking decisions. This model processes sensor data, predicts object trajectories, and handles occlusion scenarios by generating probable trajectory continuations, thereby improving both tracking accuracy and reliability without requiring direct processing of raw sensor data during occlusion events.
2Measurement precision
If complex machine-learned models are used to improve trajectory association, then tracking accuracy improves, but computational complexity increases
Solution Approach 1:
The computational system is segmented into multiple specialized components: a machine-learned model for trajectory prediction, an occlusion handling module, and a trajectory association module. Each component performs a specific function, allowing the complex overall task to be divided into manageable parts that can be processed independently and efficiently, reducing the computational burden on any single component.
Solution Approach 2:
The system applies partial action by focusing computational resources on predicting and tracking only those objects that are relevant to the autonomous vehicle's path planning. Rather than tracking all detected objects equally, the system prioritizes objects in critical zones, performing excessive computation only where necessary for safety-critical decisions, thereby balancing accuracy with computational efficiency.
3Measurement precision
If frequent trajectory updates are performed to maintain accuracy, then tracking precision improves, but processing time increases
Solution Approach 1:
The system implements periodic action by updating trajectories at optimized intervals rather than continuously. The machine-learned model predicts trajectories for multiple future time steps in batch operations, allowing the system to maintain accurate tracking information without performing full processing cycles at every possible moment. This periodic updating approach balances precision requirements with processing time constraints.
Data Source
AI summary
Systems, device, and methods for trajectory association and tracking are provided. A method can include obtaining input data indicative of a respective trajectory for each of one or more first objects for a first time step and input data indicative of a respective trajectory for each of one or more second objects for a second time step subsequent to the first time step. The method can include generating, using a machine-learned model, a temporally-consistent trajectory for at least one of the one or more first objects or the one or more second objects based at least in part on the input data and determining a third predicted trajectory for the at least one of the one or more first objects or the one or more second objects for at least the second time step based at least in part on the temporally-consistent trajectory.