Automatic 3D LIDAR Data Labeling via GPS Object Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for training autonomous vehicles to recognize objects using LIDAR data are time-consuming and prone to errors due to the manual labeling of 3D LIDAR data points, which requires significant human intervention and may not accurately identify obstacles.
Innovation Solution
A system that automatically labels 3D LIDAR data points by using GPS information from objects equipped with transceivers and wireless transmitters, allowing the autonomous vehicle to associate and label data points based on object types without user intervention, utilizing a data analytics system to perform offline association and labeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling of 3D LIDAR data points is used, then object recognition training data can be obtained, but the process consumes huge amount of time and is error prone
Solution Approach 1:
The system enables self-service by having objects automatically transmit their own GPS data and type information through wireless transmitters. Each object serves itself by providing the labeling information without requiring human annotators to manually identify and label objects in the LIDAR point cloud, thus eliminating time consumption and human error while maintaining high accuracy.
2Quantity of substance
If manual labeling of LIDAR data points is used, then training data can be collected, but the process is error prone due to difficulty in recognizing obstacles on 3D LIDAR images
Solution Approach 1:
The patent introduces GPS data and wireless transmission as an intermediary mechanism between the physical objects and the LIDAR data labeling process. Objects equipped with GPS transmitters automatically provide their location and type information, which serves as a reliable intermediary that directly links real-world objects to corresponding LIDAR data points, eliminating the need for manual interpretation of complex 3D LIDAR images and ensuring high labeling reliability.
3Productivity
If automatic labeling with GPS and wireless transmitters is used, then labeling time is reduced and accuracy is improved, but device complexity increases
Solution Approach 1:
The system achieves multi-functionality by integrating GPS transceivers and wireless transmitters into objects, allowing these components to serve multiple purposes: providing location information for data association, transmitting object type information for labeling, and enabling automatic identification without requiring separate manual processes. This universal approach increases productivity while the added complexity is distributed across individual objects rather than centralized in the processing system.
Data Source
AI summary
In one embodiment, 3D LIDAR data points are collected using a 3D LIDAR device mounted on an ADV, while the ADV is driving within a predetermined proximity. GPS information associated with a number of objects that are located and moving within the proximity surrounding the ADV. The GPS information of the objects may include a location, a speed, and a heading direction of the objects captured at a particular point in time. The objects are associated with at least some of the LIDAR data points based on the GPS information of the objects. The 3D LIDAR data points are then labeled based on a type of the objects, wherein the labeled 3D LIDAR data points are utilized to train a machine-learning algorithm or model to be utilized for object recognition by ADVs.