Autonomous Vehicle Self-Learning Framework with Logical Switching Layer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic driving algorithms struggle to adapt to complex and unknown scenarios due to their reliance on rule-based designs, limiting their ability to handle infinite real-world scenarios safely and efficiently, and existing self-evolution methods are not fully integrated with advanced AI and machine learning technologies.
Innovation Solution
A closed-loop online self-learning framework for autonomous vehicles, comprising five data closed loops: Over-the-Air Technology, online learning, algorithm evolution, self-adversarial improvement, and cloud coevolution, which manage the self-evolution process through a logical switching layer, enabling continuous adaptation and improvement of the driving algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rule-based algorithms are used for automatic driving, then the framework is clearer and more reliable, but it is very difficult to cover most automatic driving operation scenarios
Solution Approach 1:
The patent segments the automatic driving system into multiple specialized modules including perception module, prediction module, planning module, and control module. Each module handles specific aspects of driving operations, allowing the system to maintain clear internal frameworks while adapting to diverse scenarios through modular composition
Solution Approach 2:
The patent implements dynamic scenario recognition and adaptive strategy selection, where the system dynamically adjusts its behavior based on real-time environmental assessment. The logical switching layer dynamically selects different closed-loop learning modes (online learning, adversarial learning, or evolution mode) based on scenario complexity and safety requirements
2Adaptability or versatility
If self-evolution algorithm is used, then the algorithm can adapt to infinite scenarios, but closed-loop online self-learning is not achieved in fast changing scenarios
Solution Approach 1:
The patent implements continuous online learning where the system continuously updates its models using real-time data from sensors and operational feedback. The closed-loop learning process continuously refines perception, prediction, and planning models during actual driving operations, ensuring the system adapts to fast-changing scenarios without interrupting normal driving functions
Solution Approach 2:
The patent establishes multiple closed-loop feedback mechanisms including online learning loops that process real-time operational data, adversarial learning loops that generate and respond to challenging scenarios, and evolution loops that progressively improve algorithm performance. These feedback loops enable the system to learn from both normal operations and edge cases
3Ease of manufacture
If typical machine learning flow is used, then model training can be conducted, but advanced artificial intelligence and automatic driving technologies cannot be fully used
Solution Approach 1:
The patent introduces a logical switching layer as an intermediary that coordinates between different learning modes (online learning, adversarial learning, and evolution modes). This switching layer manages the complexity of multiple AI technologies by selectively activating appropriate learning mechanisms based on current operational needs and scenario characteristics
Solution Approach 2:
The patent combines multiple AI technologies and learning approaches into a composite system that integrates traditional machine learning with advanced techniques such as adversarial learning, online learning, and evolutionary algorithms. This composite approach leverages the strengths of each method while mitigating their individual limitations
Data Source
AI summary
The present invention provides a closed-loop online self-learning framework applied to an autonomous vehicle, and belongs to the technical field of automatic driving. The closed-loop online self-learning framework includes five data closed loop links, including: an Over-the-Air Technology (OTA) closed loop, an online learning closed loop, an algorithm evolution closed loop, a self-adversarial improvement closed loop, and a cloud coevolution closed loop. According to current characteristics of a self-evolution process of an algorithm, the five data closed loop links of the present disclosure are subjected to overall management through a logical switching layer of an upper layer, so as to separate a self-evolution algorithm from a typical machine learning flow, and closed-loop online self-learning of an automatic driving algorithm is achieved under a rapidly changing scenario by fully using an advanced artificial intelligence and automatic driving technology, so as to finally achieve closed-loop evolution of an automatic driving algorithm.