Multi-Sensor Lane Verification for Stable Real-Time Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing lane detection systems for autonomous vehicles face challenges in achieving precise and stable localization, especially when using differential GPS, and require improved methods for real-time detection of lane markings.
Innovation Solution
A method and system for lane detection that involves generating a ground truth for lane markings, training a confidence module to assess the reasonableness of fitted lane markings, and using a combination of sensors like GPS, IMU, and LiDAR to optimize lane template fitting and detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If differential GPS is used for localization, then the system becomes affordable, but the precision of localization is not desirably stable
Solution Approach 1:
The patent combines multiple sensors (camera, LiDAR, IMU, GPS) into an integrated sensing system. The camera and LiDAR capture visual and depth information of lane markings, the IMU provides motion data, and GPS provides location data. By merging these sensor inputs, the system achieves stable and precise localization that compensates for the instability of GPS alone.
Solution Approach 2:
The patent introduces an intermediary processing system that fuses data from multiple sensors. The confidence module acts as an intermediary that evaluates the reliability of detected lane markings by comparing multiple sensor inputs and historical data, thereby stabilizing the localization precision without requiring expensive hardware.
2Measurement precision
If multiple sensors are used for lane detection, then the precision and reliability of detection is improved, but the device complexity increases
Solution Approach 1:
The patent designs a multi-functional processing system that handles multiple sensor types (camera, LiDAR, IMU, GPS) through a unified architecture. The same processing modules (hit-map generation, lane marking detection, confidence evaluation) process data from different sensors, reducing the need for separate processing chains and minimizing overall system complexity.
Solution Approach 2:
The confidence module performs self-evaluation of detected lane markings by analyzing the consistency of multiple sensor inputs and historical detection results. This self-service mechanism automatically filters unreliable detections without requiring external validation, simplifying the system architecture while maintaining high precision.
3Productivity
If real-time lane detection is implemented, then the detection speed is improved, but the measurement precision may be compromised
Solution Approach 1:
The patent pre-generates hit-maps from historical lane detection data and stores them for quick reference. When detecting lane markings in real-time, the system compares current sensor data against these pre-processed hit-maps, significantly reducing computation time while maintaining detection accuracy through the confidence evaluation module.
Solution Approach 2:
The system dynamically adjusts the detection process by using the confidence module to evaluate each detected lane marking in real-time. Based on the confidence score, the system can quickly accept or reject detections, enabling real-time response while ensuring precision through adaptive validation rather than fixed processing thresholds.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method of lane detection for a non-transitory computer readable storage medium storing one or more programs is disclosed. The one or more programs include instructions, which when executed by a computing device, cause the computing device to perform the following steps comprising: generating a ground truth associated with lane markings expressed in god's view; receiving features from at least one of a hit-map image and a fitted lane marking, wherein the hit-map image includes a classification of pixels that hit a lane marking, and the fitted lane marking includes pixels optimized based on the hit-map image; and training a confidence module based on the features and the ground truth, the confidence module configured to determine on-line whether a fitted lane marking is reasonable, using parameters that express a lane marking in an arc.