Autonomous Driving Prediction Split for Safety and Target Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for predicting the motion trajectory of objects in relation to a host vehicle often compromise between safety and performance, either prioritizing safety at the expense of performance or vice versa, leading to potential risks or impaired target performance.
Innovation Solution
A processing method that includes a performance achievement prediction to plan the host vehicle's driving for target performance and a safety assurance prediction to independently ensure reasonably foreseeable safety, allowing for a balance between performance and safety through separate predictions and monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single prediction model prioritizes safety, then safety is improved, but target performance is impaired
Solution Approach 1:
The patent divides the prediction function into two independent prediction models: a first prediction model for safety assurance and a second prediction model for performance achievement. This segmentation allows each model to be optimized for its specific purpose without compromising the other, resolving the contradiction between safety and performance by eliminating the need to choose one over the other in a single model.
2Productivity
If a single prediction model prioritizes target performance, then productivity is improved, but safety risks increase
Solution Approach 1:
The patent segments the prediction system into two independent models where the first model focuses on safety assurance and the second model focuses on performance achievement. This allows the performance model to optimize for productivity without being constrained by safety considerations, while the safety model independently monitors and ensures safety requirements are met.
3Productivity
If separate prediction models are used for safety and performance, then both safety and performance can be optimized, but system complexity increases
Solution Approach 1:
The patent applies segmentation by creating two independent prediction models that operate separately but complement each other. This segmentation resolves the complexity issue by providing clear functional separation, making the system more manageable and easier to validate compared to a single complex model trying to balance both safety and performance simultaneously.
Solution Approach 2:
The patent introduces an integration mechanism that acts as an intermediary between the two prediction models. This intermediary coordinates the outputs of the safety prediction model and performance prediction model, combining their results to generate the final driving control decisions. This mediator structure manages the complexity by providing a systematic way to integrate the two separate models without requiring complex interactions between them.
Data Source
AI summary
A processing method, which is executed by a processor for performing a processing related to a driving of a host moving object, includes a performance achievement prediction that predicts a future action of an other road user in an external environment of the host moving object as a prediction for achieving a target performance of the host moving object; a driving plan that plans the driving of the host moving object according to the performance achievement prediction; a safety assurance prediction that predicts, independently of the performance achievement prediction, the future action of the other road user in the external environment as a prediction for assuring a reasonably foreseeable safety of the host moving object; and a driving monitoring that monitors the driving of the host moving object according to the safety assurance prediction.