Autonomous Vehicle Action Planning With Macro-Micro Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle systems face computational inefficiencies and operational slowness due to generalized object and feature detection processes, which can be improved for enhanced computational efficiency and robustness.
Innovation Solution
Implementing a 'macro-micro scheme' for vehicle operation, where a general action module determines a high-level plan (macro) and detailed action modules execute specific actions (micro), with an estimation of success rate for fulfilling the plan, optimizing each module for specific conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a generalized object and feature detection process is used for navigation and obstacle avoidance, then the vehicle operation is reliable, but the computational efficiency is low and execution is slow
Solution Approach 1:
The patent segments the action planning system into a general action module that handles high-level navigation decisions and multiple specialized detailed action modules that handle specific tasks like obstacle avoidance, lane keeping, and intersection navigation. This segmentation allows each module to be optimized for its specific function, improving overall computational efficiency while maintaining reliability through specialized handling of different driving scenarios
Solution Approach 2:
The system dynamically changes operational parameters by selecting different detailed action modules based on the current driving context. When the general action module determines a specific navigation task is required, it activates the corresponding specialized detailed action module, allowing the system to adapt its processing approach to match the specific situation, thereby improving computational efficiency without sacrificing reliability
2Reliability
If a generalized object and feature detection process is used for navigation and obstacle avoidance, then the vehicle operation is reliable, but the execution time is increased
Solution Approach 1:
The patent implements preliminary action by pre-configuring multiple detailed action modules for different driving scenarios before execution is needed. When a specific navigation task is identified by the general action module, the corresponding pre-prepared detailed action module is immediately activated, eliminating the need for runtime decision-making about which approach to use. This reduces execution time while maintaining reliability through proven, pre-tested action sequences
Solution Approach 2:
By segmenting the navigation system into specialized modules, each detailed action module can be independently optimized and tested for its specific function. This allows the system to execute only the necessary module for the current situation rather than running a complete generalized detection and planning process, significantly reducing execution time while maintaining reliable operation through specialized handling
Data Source
AI summary
The present disclosure relates to a computer implemented method for operating an autonomous vehicle based on sensor data representative of an area in a driving direction of and in the vicinity of the vehicle. The vehicle is equipped with a control unit adapted to determine if a plurality of detailed actions to be performed by the vehicle successfully may be used for fulfilling a desired general action plan for the vehicle. The present disclosure also relates to a corresponding control system and to a computer program product.