Autonomous Driving Mode Switching via Trajectory Collision Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving technologies lack effective methods for efficiently switching from manual to autonomous driving modes, especially when a driver is driving recklessly, and fail to assess collision probabilities based on recent driving routes.
Innovation Solution
A learning method and device that analyze a driver's recent driving route by using a combination of status, trajectory, and safety networks to calculate collision probabilities, determining whether to switch the vehicle's mode from manual to autonomous driving based on these analyses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual driving mode is used, then driver control is maintained, but collision risk increases when driver drives recklessly
Solution Approach 1:
The patent introduces an autonomous driving system as an intermediary between the reckless driver and the vehicle control. The system monitors driver behavior through multiple sensors and trajectory analysis, and when dangerous driving is detected, it takes over vehicle control to prevent collisions, thus resolving the contradiction between maintaining driver control and preventing collision risk
Solution Approach 2:
The system performs preliminary analysis of driver behavior patterns and potential collision risks before actual collisions occur. By continuously monitoring trajectory data and predicting future collision probabilities, the system can proactively switch to autonomous mode before dangerous situations escalate, preventing rather than just reacting to hazards
2Reliability
If mode switching is implemented, then autonomous driving can prevent reckless driving accidents, but determination of when to switch is difficult
Solution Approach 1:
The system implements continuous feedback loops that monitor driver behavior, assess collision risks in real-time, and adjust the decision to switch modes accordingly. Multiple sensors provide ongoing feedback about vehicle state and environment, while trajectory analysis provides feedback about driver intent, enabling the system to determine the optimal timing for mode switching based on accumulated evidence rather than arbitrary thresholds
Solution Approach 2:
The patent replaces subjective human judgment about when to switch modes with an automated computational system. Instead of relying on drivers or operators to decide when autonomous mode should activate, the system uses algorithmic analysis of trajectory data and collision probability calculations to objectively determine switching timing, eliminating the difficulty of human judgment in this context
3Measurement precision
If trajectory-based behavior analysis is performed, then collision probability can be calculated, but computational complexity increases
Solution Approach 1:
The patent segments the complex task of collision probability calculation into distinct computational modules: trajectory data collection from sensors, trajectory smoothing and filtering, behavior pattern recognition, collision risk assessment, and mode switching decision. This segmentation allows each module to be optimized independently and processed in sequence, reducing overall computational complexity while maintaining measurement precision
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A learning method for calculating collision probability, to be used for determining whether it is appropriate or not to switch driving modes of a vehicle capable of an autonomous driving, by analyzing a recent driving route of a driver is provided. And the method includes steps of: (a) a learning device, on condition that a status vector and a trajectory vector are acquired, performing processes of (i) instructing a status network to generate a status feature map and (ii) instructing a trajectory network to generate a trajectory feature map; (b) the learning device instructing a safety network to calculate a predicted collision probability representing a predicted probability of an accident occurrence; and (c) the learning device instructing a loss layer to generate a loss by referring to the predicted collision probability and a GT collision probability, which have been acquired beforehand, to learn at least part of parameters.