Autonomous Vehicle Localization Restart Using Stored LiDAR Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Localizing autonomous vehicles after their computing devices have been suspended is time-consuming and costly, especially when scaled across a fleet, as existing methods require manual driving to reset biases and can be prone to errors.
Innovation Solution
The method involves storing a complete LIDAR sensor scan when the vehicle is parked and comparing it with a new scan upon activation to determine if the vehicle has moved, using algorithms like ICP to filter and align data points, and utilizing a remote localization device in dynamic environments to apply transforms to stored localization data for accurate positioning and orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual driving is used to determine localization and biases, then accurate localization can be obtained, but the process becomes time-consuming and costly when scaled across a fleet of vehicles
Solution Approach 1:
The system performs preliminary localization and bias calibration during the vehicle's normal operation before suspension. Localization data and sensor biases are captured and stored in advance, so when the computing device is suspended and restarted, the pre-collected data can be quickly applied without requiring time-consuming manual driving procedures again.
Solution Approach 2:
The system creates a digital copy of the vehicle's environmental perception through LIDAR scan data and localization information. This digital twin is stored and can be rapidly compared with new scans after suspension, replacing the need for physical manual driving to re-establish localization. The copy allows for quick verification of whether the vehicle has moved by comparing point cloud data.
2Measurement precision
If miles of manual driving are required to capture sensor data and compare to map information, then accurate localization and biases are obtained, but the process becomes difficult and costly when applied to multiple vehicles
Solution Approach 1:
The vehicle performs self-localization and self-calibration using its own LIDAR sensor and onboard computing resources. The system automatically captures LIDAR scans, processes the point cloud data, compares it with stored environmental maps, and determines localization and sensor biases without requiring external human operators to manually drive each vehicle through calibration routines.
Solution Approach 2:
The system replaces the mechanical process of manual driving with an automated computational process. Instead of physically driving the vehicle to collect data, the onboard LIDAR sensor automatically captures environmental scans, and algorithms process this data to determine localization and biases, significantly improving fleet-wide productivity.
3Measurement precision
If LIDAR sensor scan data is compared to determine whether the vehicle has moved, then localization accuracy is improved, but data processing complexity increases
Solution Approach 1:
The system extracts only the essential features from the LIDAR point cloud data that are necessary for localization determination. Rather than processing every single point, the system identifies key geometric features and characteristics of the environment that can reliably indicate whether the vehicle has moved, reducing computational complexity while maintaining accuracy.
Solution Approach 2:
The system performs a simplified comparison that checks whether the vehicle has moved beyond a certain threshold of movement. Rather than calculating precise localization continuously, the system performs a partial action by determining only whether movement has occurred, which is sufficient for the suspension-resumption scenario. This reduces processing complexity while still providing accurate enough localization for resuming operation.
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
This approach reduces the time and resources needed for vehicle localization, enhancing efficiency and accuracy, and allows autonomous vehicles to quickly return to service, particularly in reducing human operator time and labor costs.
Implementation Method 1
using a LIDAR sensor to capture the first LIDAR sensor scan data and to capture the second LIDAR sensor scan data
Data Source
AI summary
The technology relates to localizing a vehicle. As one approach, a first LIDAR sensor scan data of an environment of the vehicle and localization data for the vehicle at a location where the first LIDAR sensor scan was captured are stored. Thereafter, the computing device is suspended and subsequently unsuspended. After the computing device is unsuspended, second LIDAR sensor scan data of the vehicle's environment is received. The first LIDAR sensor scan data is compared to the second LIDAR sensor scan data to determine whether the vehicle has moved. Based on the determination of whether the vehicle has moved from the location, the stored localization data is used to localize the vehicle. Other approaches are also described.