Grid-Based Trajectory Verification for Unpredictable Road Users
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle systems struggle to accurately predict the future motion of other road users, especially in scenarios where their behavior is unpredictable or does not follow established rules.
Innovation Solution
A method involving the use of a grid-based approach to predict the possible future locations of road users, generating a heat map based on sensor data, and comparing predicted trajectories to the heat map to validate their accuracy and generate new trajectories if necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional behavior prediction models are used to evaluate trajectory likelihood, then the system can process predictions efficiently, but the accuracy of predicting unpredictable road user behavior deteriorates
Solution Approach 1:
The patent segments the prediction task into two distinct components: a traditional behavior prediction model that generates candidate trajectories efficiently, and a grid-based verification system that validates these trajectories against learned motion patterns. This segmentation allows the system to maintain processing efficiency while improving accuracy through specialized verification of each trajectory segment against the heat map representation of probable locations.
Solution Approach 2:
The heat map serves as an intermediary data structure that bridges the traditional prediction model and the verification process. By pre-computing probable locations into a grid-based heat map, the system creates an intermediate representation that enables fast verification of trajectory plausibility without requiring complex real-time simulations, thus improving accuracy while maintaining efficiency.
2Measurement precision
If a grid-based verification system is implemented to validate predicted trajectories, then the accuracy of short-term predictions is improved, but the computational complexity of the system increases
Solution Approach 1:
The system performs preliminary action by pre-computing the heat map of probable locations based on historical sensor data and motion patterns before trajectory verification is needed. This pre-computation stores the results in a grid-based structure that can be quickly queried during verification, avoiding the need for complex real-time simulations and reducing overall computational complexity despite the added verification step.
Solution Approach 2:
The patent uses a computationally inexpensive grid-based heat map structure that can be quickly generated and discarded. Rather than implementing complex, long-lived prediction models, the system creates simple probabilistic grids that are regenerated as needed, providing accurate verification with minimal computational overhead and allowing the system to trade off between refresh frequency and accuracy.
3Reliability
If the system generates and compares multiple predicted trajectories against a heat map, then the reliability of trajectory validation is improved, but the processing time increases
Solution Approach 1:
The system applies partial action by verifying trajectories against only the most relevant portions of the heat map rather than performing exhaustive comparisons across the entire grid. The verification process focuses on checking whether trajectory points fall within high-probability regions identified by the heat map, allowing the system to achieve reliable validation without processing every possible trajectory variant in full detail.
Data Source
AI summary
Aspects of the disclosure provide for controlling a vehicle in an autonomous driving mode. For instance, sensor data for an object as well as a plurality of predicted trajectories may be received. Each predicted trajectory may represent a plurality of possible future locations for the object. A grid including a plurality of cells, each being associated with a geographic area, may be generated. Probabilities that the object will enter the geographic area associated with each of the plurality of cells over a period of time into the future may be determined based on the sensor data in order to generate a heat map. One or more of the plurality of predicted trajectories may be compared to the heat map. The vehicle may be controlled in the autonomous driving mode based on the comparison.