LiDAR Degradation Detection for Contamination-Aware Autonomous Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
LiDAR sensors in self-driving cars face degradation due to weather conditions and contamination, leading to inaccurate 3D map generation and potential collisions.
Innovation Solution
A machine-learning algorithm (MLA) is trained to determine the degree of LiDAR data degradation by analyzing coverage statistical data from LiDAR points and camera images, allowing for the identification of contamination levels and triggering preventive actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If LiDAR sensor operates in adverse weather conditions or with contaminated exit window, then the LiDAR sensor can still output light beams and detect reflected light, but the ability to detect objects is weakened causing degradation of LiDAR image data
Solution Approach 1:
The system performs preliminary assessment of LiDAR data quality by analyzing point cloud density and distribution patterns before using the data for critical decisions. This early detection allows the system to identify degradation caused by weather or contamination and take preventive actions or switch to alternative sensing modes before reliability is severely compromised.
2Measurement precision
If the sampling rate over the field of view is increased to generate more detailed and accurate 3D map, then the detail and accuracy of the 3D map is improved, but the data processing complexity and computational load increase
Solution Approach 1:
Instead of uniformly increasing sampling rate across the entire field of view, the system dynamically adjusts sampling density based on local importance. Regions containing critical objects or areas with higher uncertainty receive higher sampling rates, while less important regions use lower sampling rates. This maintains 3D map accuracy for critical elements while reducing overall computational complexity.
Solution Approach 2:
The system applies excessive sampling (higher point cloud density) only in specific regions where it is most needed, such as areas with detected objects or regions with poor visibility, rather than uniformly across the entire field of view. This partial application of high sampling rates achieves necessary measurement precision while limiting the increase in data processing complexity.
3Reliability
If the LiDAR sensor field of view is obscured by contamination or weather conditions, then the detection accuracy is compromised, but continuing operation without intervention may result in collisions
Solution Approach 1:
The system continuously monitors LiDAR data quality metrics including point cloud density, signal strength, and detection consistency. When degradation is detected, the system provides feedback to the vehicle control system to adjust operations, such as reducing speed, increasing following distance, or switching to alternative sensing modalities, thereby maintaining safety despite reduced detection accuracy.
Solution Approach 2:
When LiDAR degradation is detected, the system proactively implements compensatory measures before critical failures occur. This includes switching to radar or camera-based detection, increasing safety margins in trajectory planning, and preparing alternative response strategies, thereby cushioning against the potential harmful effect of collisions even when detection accuracy is compromised.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The MLA effectively increases the accuracy of LiDAR data degradation assessment, enabling timely preventive actions such as cleaning the LiDAR sensor or adjusting the vehicle's trajectory, thereby reducing the risk of collisions.
Implementation Method 1
LiDAR sensors allow measurement of distances between the system and surrounding objects by irradiating the surroundings with light and collecting light reflected from one or more objects
Implementation Method 2
For Time-of-Flight (ToF) systems, for example, time between (i) emission of the light beam and (ii) detection of the reflected and/or scattered light beam is measured to determine the distance to the object
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for for determining a degree of point cloud data degradation of a LiDAR sensor of a Self-Driving Car (SDC) using a machine-learning algorithm (MLA) are provided. The method comprises: determining, based on a training point cloud generated by the LiDAR sensor representative of surroundings of the SDC, a plurality of LiDAR features; determining, for each training object in the surroundings, based on statistical data of coverage of training objects with LiDAR points, a plurality of enrichment features; receiving a respective label indicative of a degradation degree of the training point cloud; generating, based on the plurality of LiDAR features, the plurality of enrichment features, and the respective label, a given feature vector of a plurality of feature vectors; training, based on the plurality of feature vectors, the MLA to determine an in-use degree of degradation of in-use sensed data further generated by the LiDAR sensor.