Dual-Model Path Control for Smoother Autonomous Driving Transitions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous driving systems experience increased user intervention due to significant changes in control instructions resulting from model changes, such as switching from rule-based to machine learning models, leading to discomfort and nonconformity.
Innovation Solution
A control device that maintains both a new and original control model, using conformability criteria adjusted by user feedback to prioritize one model over the other based on user intervention, generating a final control path that minimizes nonconformity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a new control model is deployed to improve autonomous driving performance, then control accuracy and intelligence are improved, but user intervention frequency increases due to significant changes in control instructions
Solution Approach 1:
The patent introduces a path generation unit that acts as an intermediary between the new and original control models. This unit generates multiple candidate paths using both models and selects or combines them based on conformability criteria, thereby mediating the transition and reducing abrupt changes that cause user intervention.
Solution Approach 2:
The patent dynamically adjusts the conformability criteria parameters based on user feedback and operational context. By changing the weighting parameters in the conformability evaluation, the system can adaptively balance between following the new model's instructions and maintaining consistency with the original model's behavior, thus reducing user intervention.
2Adaptability or versatility
If control instructions change significantly due to model updates, then control intelligence is improved, but automatic control nonconformity increases
Solution Approach 1:
The patent performs preliminary evaluation of candidate paths against conformability criteria before final selection. By预先 assessing the conformability of paths generated by the new model and comparing them with paths from the original model, the system prevents significant nonconformity from reaching the execution stage.
Solution Approach 2:
The patent implements a feedback mechanism where user interventions are recorded and used to adjust the conformability criteria. This feedback loop allows the system to learn from nonconformity instances and adapt its path selection strategy, thereby reducing future automatic control nonconformity while maintaining improved control intelligence.
3Productivity
If the new control model is prioritized to maximize performance benefits, then control effectiveness is improved, but user discomfort increases due to frequent interventions
Solution Approach 1:
The patent dynamically adjusts the prioritization between the new and original control models based on real-time conditions and user feedback. The conformability criteria weights are not fixed but adapt according to the situation, allowing the system to maximize control effectiveness when the new model performs well while reducing user discomfort when interventions are needed.
Data Source
AI summary
The control device according to one aspect of the present disclosure generates a first path and a second path using a first control model and a second control model, evaluates the deviation between the generated first path and second path using a compatibility criteria, generates a final path from the first path and the second path by reflecting the evaluation result, and controls the movement of a moving body in accordance with the generated final path. The conformability criteria have been adjusted by user feedback so that the less an intervention operation by the user is performed, the more the first path is prioritized in generate the final path, and the more the intervention operation by the user is performed, the more the second path is prioritized in generate the final path.