Remote Assistance Operator Matching by Scene Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Operators assigned to handle remote assistance requests from autonomous driving vehicles face increased burden due to the need to adapt to diverse traffic environments and tasks, leading to fatigue and inefficiency.
Innovation Solution
A system that utilizes a database to manage feature values of past assistance scenes for operators, allowing for the selection of candidate operators whose past scenes are similar to current requests, thereby reducing the burden by assigning operators with relevant experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an operator is randomly selected each time for remote assistance requests, then operator availability is improved, but operator burden increases due to needing to adapt to diverse traffic environments and tasks
Solution Approach 1:
The system performs preliminary classification of assistance scenes into multiple categories (e.g., traffic jam, accident, natural disaster) and pre-assigns operators who are familiar with specific scene types. This preliminary organization allows operators to be matched with appropriate requests based on their expertise, reducing the need to adapt to diverse environments and lowering operational burden while maintaining availability through structured assignment
2Ease of operation
If operators continuously support one autonomous driving vehicle, then operator burden is reduced, but operator availability decreases due to inability to handle multiple requests
Solution Approach 1:
The system segments the pool of operators into multiple groups based on their expertise in different assistance scene categories. Each operator is assigned to handle requests within their specialized domain, allowing them to continuously support vehicles in similar scenarios without being overwhelmed by diversity. This segmentation enables operators to maintain lower burden while the system as a whole maintains high availability through multiple specialized operator groups
Solution Approach 2:
The system creates a universal assignment mechanism that can dynamically allocate operators across different vehicle requests based on scene category matching. Operators maintain specialized knowledge for specific scene types but can universally handle multiple requests within their domain, achieving both continuous support for similar scenarios and availability for multiple assignments
3Ease of operation
If operators are assigned based on similar past experience, then operator burden is reduced, but system complexity increases due to need for scene feature management and similarity determination
Solution Approach 1:
The system transforms the complex problem of operator assignment by changing the parameter representation from raw scene data to extracted feature values. By identifying and managing only the key feature items relevant to scene similarity (e.g., scene type, location characteristics, task category), the system reduces dimensionality and complexity while still enabling effective matching based on similar past experience
Data Source
AI summary
A remote assistance system according to the present disclosure comprises a memory storing a database and one or more processors. The database manages, for each of a plurality of operators, feature values of a plurality of feature items regarding a last assistance scene. The one or more processors are configured to execute the following first to third processes. The first process is, when a new assistance request is received from s vehicle, specifying one or more index items from the plurality of feature items. The second process is specifying, from the plurality of operators, one or more candidate operators whose the last assistance scene is similar to a current assistance scene based on the feature values of the one or more index items. The third process is selecting an operator to process the new assistance request from the one or more candidate operators.