Monitoring Server Distributed Processing for Robot Image Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing distributed processing systems fail to effectively manage image analysis processes between apparatuses with different processing capabilities, leading to excessive loads on smaller devices or robots, and inability to perform processes in real-time augmented reality services.
Innovation Solution
A monitoring server with a communication unit and distributed-processing control unit that determines which processes to perform on the information processing apparatus and the monitoring server based on the type of required output for image analysis, considering communication quality and processing load information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If image analysis processes are distributed to apparatuses with different processing capabilities, then processing load is reduced on individual devices, but smaller devices cannot perform the same processes as server apparatus
Solution Approach 1:
The image analysis process is divided into multiple stages: first process (e.g., face detection) that can be performed on smaller devices with limited capabilities, and second process (e.g., gender recognition, number counting) that is performed on server apparatus with higher capabilities. This segmentation allows each apparatus to perform only the processes it can handle, resolving the contradiction between productivity and adaptability.
Solution Approach 2:
The system dynamically determines which apparatus performs which process based on the type of output required. When only face detection is needed, the small device performs the entire process. When additional analysis like gender recognition is required, the system transitions to using the server apparatus for the second process, making the distribution strategy flexible and adaptive to different requirements.
2Speed
If real-time image processes are distributed to small-sized devices, then processing can be performed locally, but excessive load is imposed on the small-sized device
Solution Approach 1:
The image analysis process is segmented into first process (face detection) and second process (gender recognition, number counting). The small-sized device performs only the first process which has lower computational requirements, while the server apparatus performs the second process. This segmentation reduces the processing load on the small-sized device while maintaining real-time performance for the critical face detection function.
Solution Approach 2:
The small-sized device performs only the necessary first process (face detection) rather than the complete image analysis pipeline. By performing only the partial process that is essential for real-time tracking and leaving the more demanding second process to the server, the system achieves real-time performance without excessive load on the small-sized device.
3Productivity
If all image analysis processes are performed on server apparatus, then processing capability is sufficient, but communication load and latency increase
Solution Approach 1:
The image analysis process is segmented such that the first process (face detection) is performed on the small-sized device and only the results are transmitted to the server apparatus for the second process. This segmentation reduces the amount of data that needs to be communicated, thereby reducing communication load and latency while still utilizing the server's superior processing capability for the more complex second process.
Solution Approach 2:
The small-sized device performs preliminary processing (face detection) before transmitting data to the server apparatus. By performing this preliminary action locally, the system reduces the volume of data that needs to be transmitted and processed by the server, thereby reducing communication latency and improving overall processing efficiency.
Data Source
AI summary
A monitoring server includes a communication unit (11) that receives image data transmitted periodically from a robot (20), information on communication quality of a communication network between the monitoring server and the robot (20), and load information of the robot (20), and a distributed-processing control unit (12) that determines, based on at least one of tracking-operation-environment information of the tracking target (30) determined based on the image data, the information on the communication quality, and the load information, a process to be performed in the robot (20), and the process to be performed in the monitoring server itself from among an image analysis process including a plurality of processes, the image analysis process being a process of specifying the tracking target (30), in which the communication unit (11) transmits information on the process to be performed in the robot (20) to the robot (20).