Delivery Robot Localization via Ad-Hoc V2X Network Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic package delivery systems face challenges in achieving high precision location estimates due to poor GPS satellite visibility, particularly in areas with tree cover and building obstructions, making them costly and non-scalable without expensive 3D sensors.
Innovation Solution
The implementation of an ad-hoc network that includes autonomous delivery vehicles, 5G mobile device transceivers, and vehicle-to-infrastructure (V2I) enabled vehicles, which communicate using millimeter-wave components to compute location estimates and steer robots for maximum positioning accuracy, eliminating the need for expensive 3D sensors by utilizing 2D occupancy maps for navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS based techniques are used for location estimation, then the system is simple and low cost, but the measurement precision is insufficient in areas with poor satellite visibility
Solution Approach 1:
The patent combines multiple localization techniques (GPS, Wi-Fi positioning, inertial measurement units, and visual odometry) into a fused estimation system. This merging allows the system to maintain high measurement precision in areas with poor satellite visibility by compensating GPS weaknesses with alternative methods, while avoiding the need for expensive standalone 3D sensors.
Solution Approach 2:
The patent implements a multi-functional localization system that can operate using different methods depending on environmental conditions. The system universally handles both GPS-based open-sky localization and sensor-based indoor/obstructed localization, making it adaptable to various scenarios without requiring specialized expensive hardware for each condition.
2Measurement precision
If expensive 3D sensors such as 3D-LiDAR are used on delivery robots, then the measurement precision and navigation capability are improved, but the cost increases and scalability is reduced
Solution Approach 1:
Instead of equipping each robot with expensive 3D sensors, the patent creates a virtual 3D environment by fusing data from multiple robots and infrastructure sensors. Each robot contributes its sensor data to build a collective understanding of the environment, allowing individual robots to navigate with high precision using only inexpensive 2D sensors.
Solution Approach 2:
The patent replaces expensive, complex 3D sensing hardware with inexpensive 2D cameras and sensors on each robot. The system compensates for the limitations of these cheap sensors through networked collaboration and data fusion, achieving high navigation precision without the cost burden of 3D-LiDAR on every unit.
3Measurement precision
If ad-hoc network with multiple vehicles and infrastructure is implemented, then the localization accuracy is improved, but the device complexity and communication requirements increase
Solution Approach 1:
The patent divides the localization task into segments performed by different system components: infrastructure sensors provide fixed reference points, moving vehicles contribute mobile sensing data, and each robot performs local processing. This segmentation allows the system to achieve high localization accuracy through distributed computation rather than requiring complex centralized processing.
Data Source
AI summary
A method for computing a quality location estimate of a delivery robot by creating an ad-hoc network that can include one or more autonomous delivery vehicles, nearby infrastructure such as 5th Generation signal transceivers, vehicle-to-infrastructure (V2I) enabled autonomous vehicles, and millimeter-wave device components in Line-Of-Sight (LOS) with any of the above communicating devices. The method can include estimating the quality for localization (e.g., dilution of precision), and steering the robot delivery vehicle via a vehicle-to-anything (V2X) antenna disposed on the robot delivery vehicle and/or repositioning the autonomous delivery vehicle itself to obtain maximum positioning accuracy. The location estimates computed by the vehicle are sent to the delivery robot which then fuses these estimates with its onboard sensor values. The method may assist localization based on a 2D occupancy map to enhance the positioning performance and provides robust localization mechanism without expensive 3D sensors.