LIDAR Waveform Filtering for Spurious Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in distinguishing between actual objects and spurious objects like vehicle exhaust, dust, rain, snow, or fog, which can be misinterpreted by LIDAR sensors, leading to inappropriate driving decisions.
Innovation Solution
A method involving the use of LIDAR data points with waveform information to train models or apply heuristics, allowing vehicles to identify and filter out spurious objects by analyzing peak elongation, number of peaks, and peak width, thereby improving the accuracy of object detection and driving decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If LIDAR sensors are used to detect objects in autonomous vehicles, then the vehicle can perceive its surroundings, but the vehicle may misidentify spurious objects (exhaust, dust, rain, snow, fog) as solid objects
Solution Approach 1:
The patent transitions from analyzing only intensity information (1D) to utilizing waveform data (temporal dimension) of LIDAR returns. By examining the temporal profile of light intensity over time, the system can distinguish spurious objects from solid objects based on their different waveform characteristics, adding a temporal dimension to the detection process.
Solution Approach 2:
The patent changes the detection parameters from simple intensity thresholds to waveform-based parameters such as peak elongation, number of peaks, and peak width. These parameter changes enable the system to differentiate between spurious objects (which exhibit specific waveform patterns) and solid objects (which have different waveform characteristics).
2Reliability
If the vehicle responds to all detected LIDAR objects, then safety is prioritized, but unnecessary responses to spurious objects reduce efficiency and increase false alarms
Solution Approach 1:
The patent extracts and filters out spurious objects from the set of detected LIDAR returns by analyzing waveform characteristics. By identifying and removing returns corresponding to spurious objects (exhaust, dust, rain, snow, fog) based on their unique waveform patterns, the system prevents unnecessary safety responses while maintaining protection against real obstacles.
Solution Approach 2:
The system uses waveform analysis feedback to continuously refine object classification. By comparing the temporal waveform patterns of detected objects against known characteristics of spurious versus solid objects, the system provides feedback to the detection algorithm, improving its ability to distinguish between the two types of objects and reduce false alarms.
3Measurement precision
If waveform data analysis is implemented to distinguish spurious objects, then detection accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments the waveform analysis into distinct, computationally manageable features such as peak elongation, number of peaks, and peak width. By breaking down the complex waveform analysis into these discrete segments, the system reduces computational complexity while maintaining the ability to accurately distinguish spurious objects from solid objects.
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
Enhances the ability of autonomous vehicles to accurately differentiate between solid and spurious objects, improving safety and efficiency by reducing unnecessary responses to environmental factors like precipitation or vehicle exhaust.
Implementation Method 1
a plurality of LIDAR data points generated by a LIDAR sensor of the vehicle, each given LIDAR data point (1) including location information and intensity information and (2) being associated with waveform data
Implementation Method 2
LIDAR data points generated by a LIDAR sensor
Data Source
AI summary
Aspects of the disclosure relate to detecting spurious objects. For instance, a model may be trained using raining data including a plurality of LIDAR data points generated by a LIDAR sensor of a vehicle. Each given LIDAR data point includes location information and intensity information, and is associated with waveform data for that given LIDAR data point. At least one of the plurality of LIDAR data points is further associated with a label identifying spurious objects through which the vehicle is able to drive. The model and/or a plurality of heuristics may then be provided to a vehicle in order to allow the vehicle to determine LIDAR data points that correspond to spurious objects. These LIDAR data points may then be filtered from sensor data, and the filtered sensor data may be used to control the vehicle in an autonomous driving mode.