Traffic Light Detection Using Semantic Maps and Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional autonomous driving algorithms are impractical for complex routes due to the need for numerous decisions, and existing systems lack effective methods for accurately detecting traffic lights and predicting traffic signal transitions, leading to potential collisions and inefficiencies.
Innovation Solution
The system uses multiple processors to detect traffic lights by combining digital video streams from cameras with sensor data from LiDARs, RADARs, and GNSS receivers, filtering data using Bayesian or Kalman filters, and referencing semantic maps to determine spatiotemporal locations and predict traffic signal transitions, allowing for safe and efficient vehicle operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional greedy algorithms are used for autonomous driving decisions, then the system is simple to implement, but it becomes impractical for complex routes requiring many decisions
Solution Approach 1:
The system performs preliminary actions by pre-detecting traffic light locations using semantic maps and sensor data fusion before reaching the intersection. The Kalman filter predicts traffic light positions and signal transitions in advance, allowing the vehicle to plan routes through complex intersections without real-time decision paralysis.
Solution Approach 2:
The detection system is segmented into specialized modules: LiDAR for 3D spatial mapping, RADAR for velocity detection, cameras for visual confirmation, and semantic maps for prior knowledge. Each sensor type processes specific aspects of traffic light detection, dividing the complex task into manageable segments that can be processed in parallel.
2Measurement precision
If multiple sensors and filters are used to accurately detect traffic lights, then detection precision is improved, but system complexity increases
Solution Approach 1:
The Kalman filter implements feedback by continuously comparing predicted traffic light positions with actual sensor measurements, adjusting predictions based on measurement residuals. This feedback loop maintains high detection accuracy while filtering out noise and false positives from the multiple sensors.
Solution Approach 2:
The semantic map serves multiple functions: it provides prior knowledge of traffic light locations, validates sensor detections, predicts upcoming intersections, and confirms traffic signal states. This multi-functional component reduces the need for separate verification systems, managing complexity while maintaining precision.
3Loss of time
If traffic signal transitions are predicted in advance, then vehicle response time is improved, but false predictions may cause unnecessary stops
Solution Approach 1:
The system performs preliminary detection of traffic light locations and signal transitions using semantic maps and Kalman filter predictions before the vehicle reaches the intersection. This advance detection allows the vehicle to prepare for signal changes while maintaining the ability to adjust based on real-time sensor confirmation.
Solution Approach 2:
The Kalman filter acts as an intermediary between raw sensor data and control decisions. It smooths noisy sensor inputs, predicts traffic light positions and signal transitions, and provides confidence estimates that help distinguish reliable predictions from false positives, reducing unnecessary stops.
4Reliability
If sensor data is filtered using Bayesian or Kalman filters, then data reliability is improved, but computational processing time increases
Solution Approach 1:
The system uses periodic action by implementing event-triggered filtering that processes sensor data at optimized intervals rather than continuously. The Kalman filter updates predictions based on predicted traffic light positions and signal transitions, processing data only when significant changes are detected or at predetermined time intervals, reducing computational overhead while maintaining reliability.
Data Source
AI summary
A method for operating a vehicle includes detecting a traffic light located at a first spatiotemporal location based on a first digital video stream captured by a first camera and a second digital video stream captured by a second camera. It is determined that the vehicle is located at a second spatiotemporal location by validating first location data received from sensors against second location data obtained by filtering the first location data. It is determined that the traffic light is expected at the first spatiotemporal location based on a semantic map referenced by the second spatiotemporal location. Responsive to determining that the traffic light is expected at the first spatiotemporal location, a traffic signal of the traffic light is detected based on the two digital video streams. A trajectory is determined in accordance with the traffic signal. A control circuit operates the vehicle in accordance with the trajectory.