Autonomous Driving Control Using Experience-Based Route Judgments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face frequent unexpected cancellations due to inability to handle various road conditions and traffic scenarios, leading to degradation of autonomous driving functionality and user experience.
Innovation Solution
An autonomous driving assistance system that utilizes a database to store and retrieve experience information for determining control methods based on sensor and map data, allowing for adaptive decision-making and reducing unexpected cancellations by learning from user-driven manual control experiences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If the autonomous driving vehicle uses a judgment function to determine control methods based on sensor information and map information, then the autonomous driving traveling can be realized, but the vehicle cannot handle all situations at the same level as user driving, leading to frequent cancellations
Solution Approach 1:
The system pre-registers experience information including status recognition information, control method judgment information, and vehicle operation information in a database before actual driving situations occur. When the autonomous driving vehicle encounters a situation, it queries the pre-registered experience information to determine control methods, avoiding the need to make judgments in real-time and reducing cancellation frequency.
Solution Approach 2:
The system establishes a feedback loop where the autonomous driving vehicle queries experience information from the database, executes control methods, and the results are used to update and refine the experience information in the database. This continuous feedback mechanism improves the reliability of the judgment function over time by learning from actual driving outcomes.
2Reliability
If the autonomous driving vehicle stops or transfers control authority when the judgment function cannot determine control methods, then safety is maintained, but the autonomous driving traveling function degrades in duration and target area
Solution Approach 1:
By pre-registering experience information for various driving situations in the database, the system enables the autonomous driving vehicle to handle more situations without stopping or transferring control. The vehicle queries the database for pre-prepared control methods, extending the duration and target area of autonomous driving traveling while maintaining safety through proven control strategies.
3Adaptability or versatility
If the system uses latest time stamp history information to determine traveling methods, then recent experiences are utilized, but the information may not reflect current situations accurately, especially on roads with small traffic volume
Solution Approach 1:
The system changes the parameter for selecting experience information from simply 'latest time stamp' to a multi-dimensional approach considering status recognition information, control method judgment information, and vehicle operation information. This allows the system to find the most relevant experience information that accurately reflects current situations, even on roads with small traffic volume, by matching multiple parameters rather than just temporal recency.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Cancelation of autonomous driving traveling which is not expected by a user is reduced. An autonomous driving assistance apparatus includes: a status recognition unit that outputs status recognition information, including status information about a status of a vehicle and a peripheral status of the vehicle; a manual control unit that outputs operation information by manual operation; an experience information database that stores the status information belonging to the status recognition information and experience information about a control method for the vehicle to travel along a route by associating the status information with the experience information; an experience information registration unit that manages the experience information database; an experience information search unit that extracts the experience information corresponding to a search key from the experience information database by using the status information as the search key; a control judgment unit that: judges whether autonomous driving traveling is possible or not, based on the experience information extracted by the experience information search unit; cancels the autonomous driving traveling if the autonomous driving traveling is impossible; and calculates a control method for the autonomous driving traveling based on the experience information and outputs control method information if the autonomous driving traveling is possible; and a control execution unit that operates the vehicle based on the control method information, wherein during a period of time after the autonomous driving traveling is canceled until the autonomous driving traveling is resumed, the experience information registration unit registers the experience information, which indicates the control method obtained from the operation information, in the experience information database by associating the experience information with the status information.