Human-Like Speed Planning for Competing and Yielding Traffic
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in making accurate speed decisions in complex scenarios due to unclear competing/yielding relationships with other traffic participants, leading to potential collisions or premature braking, which affects traffic efficiency and user experience.
Innovation Solution
A speed recommendation method that utilizes fused features and lateral decision information from human driving behavior data to provide human-like speed planning, enhancing traffic efficiency and safety by determining recommended speeds and probabilities based on driving scenarios and potential collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If acceleration is performed to pass a potential conflict obstacle, then traffic efficiency is improved, but collision risk increases when the obstacle's real intention conflicts with the planned path
Solution Approach 1:
The system performs preliminary analysis of the obstacle's behavior patterns and potential intentions before making speed decisions. By pre-processing driving data to identify competing/yielding relationships and potential conflicts, the system prepares multiple speed recommendation options in advance, allowing for safer and more efficient decision-making when conflicts are detected.
2Reliability
If direct deceleration is selected to yield to a potential conflict obstacle, then collision risk is reduced, but traffic efficiency and user experience deteriorate due to premature braking
Solution Approach 1:
The system dynamically adjusts speed recommendations based on real-time analysis of the obstacle's actual behavior and changing traffic conditions. Instead of fixed deceleration strategies, the system continuously updates speed advice by fusing multiple data sources and re-evaluating conflict probability, allowing for optimal balance between safety and efficiency as situations evolve.
Solution Approach 2:
The system changes key parameters such as speed recommendation, acceleration rate, and decision thresholds based on the analyzed competing/yielding relationships. By adjusting these parameters dynamically according to the obstacle's detected intentions and environmental context, the system avoids premature braking while maintaining collision avoidance capabilities.
3Device complexity
If the competing/yielding relationship with obstacles is not clearly identified, then speed decision complexity is reduced, but decision accuracy and safety deteriorate
Solution Approach 1:
The system introduces an intermediary analysis layer that processes driving data to explicitly identify competing/yielding relationships between the ego vehicle and obstacles. This intermediary module analyzes obstacle behaviors, predicts intentions, and structures unstructured traffic scenario data into clear relational models, making complex decision-making transparent and accurate without requiring direct complex processing in the speed decision module.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
Embodiments of this application provide a speed recommendation method and a related device, and relate to the driving field. In the method, a fused feature and lateral decision information of a first vehicle are obtained by using driving data of the first vehicle in a first time period, to determine speed recommendation information of the first vehicle based on the fused feature and the lateral decision information. Because the driving data of the first vehicle is human driving behavior data, the speed recommendation method in embodiments of this application may provide a human-like speed planning service, to obtain the speed recommendation information of the first vehicle, and the speed recommendation information may be used to implement human-like driving control. This effectively helps improve traffic efficiency, and can also ensure driving safety.