요약 : 클라우드 서버를 이용해서 로봇의 computing / storage resources 이상의 성능을 발휘할 수 있다.
1. INTRODUCTION
개발 현황
- Davinci project
- robot algorithm is implemented in the Map/Reduce
- ROS is chosen as the messaging frame.
- RoboEarth
- search engine in the field of robots.
- Rapyuta is an open source cloud frame based on the RoboEarth.
- Linux containers for service security.
- ROS nodes are running in the computing environment.
- micROS-cloud
- Vision API of Google Cloud Platform
- NOOS
However, those cloud services are from different providers and their access interfaces are quite different, which causes significant difficulty for task integration.
CloudROS
- The cloud provides computing and storing services in form of ROS nodes.
- The cloud also converts the third-party web services to ROS compatible cloud services.
- The cloud exposes its REST API to users and robots.
2. CLOUDROS FRAMEWORK
A. ROS
B. REST API
- non-ROS interfaces are required to assist the dynamic ROS network establishing task.
- The design style of a RESTful architecture
- The Web resource is the core of REST architecture.
- It realizes a loose coupling between the resource and view functions.
- The standart HTTP request methods are used to request the web service.
- It follows the C/S framework.
- URI
- A typical URI for computing service : http://server_ip:port/compute/<service>/<action>/<token>
- server_ip:port : IP address of the cloud server and the port that the clod service manager continuously listens to.
- compte : computing service is requested.
- service : cloud computing services provided by the cloud robotics platform.
- action : particular operation that controls status of the services.
- token : unique security credentials.
- A typical URI for storage service : http://server_ip:port/storage/<robotID>/<action>/<token>
- storage : cloud storing service is requested.
- robotID : unique ID of the robots.
- action : operations on the cloud data sets, include "fetch" and "store".
- A typical URI for computing service : http://server_ip:port/compute/<service>/<action>/<token>
C. Docker
- Docker container technique is used as the computing environment for the cloud system to ensure services isolation and system security.
- lighter and start-up speed is faseter.
- It needs to install the ROS and deploy corresponding ROS packages and configure the command sets for managing service nodes.
D. Architecture of Cloud System

Robot-Cloud
- All cloud service nodes can be executed in the dock container which is pre-installed with ROS environment.
- Based on the non-ROS interface protocols(?REST API?), URI of the ROS master node can be transmitted to the cloud service manager and the cloud service nodes can connect and join the ROS network located at the local robot.
- robot and cloud have a common ROS master nodem the messages can be automatically transferred based on ROS topics.
Cloud-User
- The user can access the public data stored in the cloud via the REST APIs.
- The monitoring applications should also be developed as a ROS network.
E. Cloud Service Nodes Integration
3. CLOUDROS ENABLED MOBILE ROBOT PLATFORM
A. Mobile Robot Platform
B. Kinematic Model
~~ pass ~~
4. EXPERIMENT RESULTS
A. Computing Service Experiment


B. Storing Service Experiment
각 로봇은 각자의 저장소에 데이터를 저장한다. 그리고 각 로봇의 저장소마다 새로운 정보가 또다시 하나의 저장소에 업데이트 된다. User는 하나의 저장소에 있는 정보를 사용한다.
'논문리뷰' 카테고리의 다른 글
클라우드 로봇 지능 플랫폼 기술 동향과 연구 소개 (0) | 2022.02.08 |
---|---|
구조구난 로봇을 제어하기 위한 분산처리 기반의 플랫폼 설계 (0) | 2021.09.08 |
로봇 소프트웨어 개발을 위한 클라우드 기반 통합 개발 환경 (0) | 2021.09.06 |