Traffic Light Decision Control Under Temporary Signal Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing proceed-or-stop determination apparatuses for self-driving vehicles struggle to stably determine whether to proceed or stop at traffic lights, especially when the lighting state cannot be identified due to shielding objects or weather conditions.
Innovation Solution
A proceed-or-stop determination apparatus equipped with a camera, speed acquisition unit, position acquisition unit, and a controller that recognizes traffic lights, determines the lighting state, and decides whether to proceed or stop based on the recognized state, traveling speed, and vehicle position, while also considering previous determination results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the lighting state of traffic light is estimated based on previous image only, then the vehicle can make a determination when current image is unavailable, but the determination stability deteriorates due to lack of real-time lighting state information
Solution Approach 1:
The system performs preliminary recognition of traffic light positioning and prepares determination logic in advance. When current image data is unavailable, the system can immediately apply pre-prepared determination rules based on previous lighting states, reducing the time gap and improving response reliability without requiring real-time image processing.
Solution Approach 2:
The determination result from previous cycles is fed back to the current cycle's decision-making process. This feedback mechanism allows the system to maintain determination stability by comparing current unavailable state with historical determination patterns, ensuring consistent vehicle control decisions even when real-time lighting state information is lost.
2Speed
If the vehicle determines proceed-or-stop based only on current lighting state recognition, then the response is immediate, but the determination becomes unstable when lighting state cannot be identified due to shielding objects or weather
Solution Approach 1:
The system performs preliminary positioning of traffic lights and pre-loads determination rules before actual determination is needed. This preparation enables immediate response when lighting state is recognizable, while also having backup determination logic ready to activate instantly when shielding or weather conditions obscure the traffic light, maintaining both speed and reliability.
Solution Approach 2:
The system cushions against potential determination failures by having redundant determination logic prepared in advance. When current lighting state recognition fails due to shielding objects or weather, the pre-prepared fallback mechanisms (using previous determination results and historical data) are immediately activated, preventing determination instability without sacrificing normal response speed.
3Productivity
If the vehicle maintains previous determination result during unknown lighting state, then unnecessary deceleration is reduced, but the vehicle may miss actual traffic light state changes
Solution Approach 1:
The system employs periodic determination cycles where it alternates between maintaining previous determination results during unknown states and performing fresh recognition when conditions improve. This periodic switching ensures travel efficiency is maintained during obscuration events while periodically verifying against actual lighting states to detect any changes, balancing productivity and detection accuracy.
Solution Approach 2:
When lighting state is temporarily unknown due to shielding or weather, the system skips the recognition step and directly maintains the previous determination result. This skipping mechanism prevents unnecessary deceleration and maintains travel efficiency during brief obscuration events, while the system remains ready to resume normal recognition cycles to detect actual traffic light state changes.
Data Source
AI summary
Proceed-or-stop determination apparatus, includes: camera mounted on self-driving vehicle and configured to acquire image in traveling direction of the vehicle; speed acquisition unit configured to acquire traveling speed of the vehicle; position acquisition unit configured to acquire position of the vehicle with respect to stop position corresponding to traffic light in the traveling direction; and controller. The controller is configured to perform: recognizing the traffic light at predetermined cycle based on the image; recognizing lighting state of the traffic light; and determining whether the vehicle should proceed or stop at the predetermined cycle based on the lighting state, traveling speed, and position, when the traffic light is recognized. The controller determines whether the vehicle should proceed or stop further based on previous determination result in previous cycle and the lighting state recognized in the previous cycle.