Autonomous Driving Collision Warning With Driver Takeover Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional advanced driver assistance systems (ADAS) fail to transfer control authority to the driver during unexpected situations in autonomous driving, exposing drivers to collision risks.
Innovation Solution
A driver assistance apparatus and method that utilizes a camera, radar, and controller to predict collisions by processing image and radar data, outputting warnings, and controlling steering, acceleration, or deceleration based on driver input to avoid collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the vehicle operates in autonomous traveling mode with collision prediction and warning systems, then collision risk is reduced through automated control, but the driver loses control authority during unexpected situations
Solution Approach 1:
The system dynamically adjusts control authority between autonomous and manual modes based on detected driver input. When the driver provides steering, acceleration, or braking input, the system transitions from autonomous control to manual control, allowing the driver to take over when needed while maintaining automated safety interventions during normal operation
Solution Approach 2:
The system continuously monitors driver input through sensors detecting steering wheel torque, accelerator pedal position, and brake pedal activation. This feedback mechanism enables the system to recognize when the driver intends to take control and automatically transfers authority accordingly, ensuring the driver remains in charge during unexpected situations
2Loss of information
If the system outputs multiple warnings with different time thresholds, then driver awareness is enhanced, but the response time for actual collision avoidance is reduced
Solution Approach 1:
The system performs preliminary warning actions by issuing first warnings at longer time thresholds (e.g., 3 seconds) before actual collision risk materializes. This allows the driver to become aware of potential risks in advance and prepare for corrective action, while the system continues to monitor and can intervene if no driver response occurs
Solution Approach 2:
The warning system is segmented into multiple levels based on time-to-collision thresholds. First warnings are issued for longer time thresholds to provide early notice, while second warnings are issued for shorter time thresholds when collision risk becomes more immediate. This segmentation allows the system to provide comprehensive information while maintaining adequate response time at each stage
3Reliability
If the controller automatically controls steering, acceleration, or deceleration to avoid collision, then collision avoidance is improved, but the driver's intended maneuver may be overridden
Solution Approach 1:
The system dynamically switches between autonomous control and manual control based on driver input detection. When the driver provides steering, acceleration, or braking input, the system transfers control authority to the driver, allowing intended maneuvers to proceed. When no driver input is detected and collision risk remains, the system maintains or initiates automated control interventions
Data Source
AI summary
Disclosed herein is a driver assistance apparatus including a camera having a field of view around a vehicle and configured to acquire image data with, a radar installed in the vehicle and having a sensing area around the vehicle to acquire radar data, and a controller configured to process the image data and the radar data, wherein the controller determines whether a collision with a nearby object of the vehicle is predicted in response to processing the image data and the radar data when a traveling mode of the vehicle is an autonomous traveling mode, outputs a warning about risk of the collision when the collision with the object is predicted, and controls at least one of a steering, an acceleration, or a deceleration according to a detected driver's manipulation when detecting the driver's manipulation for at least one of the steering, the acceleration, or the deceleration after outputting the warning.