leoregner / object-recongition

Master's Thesis

Start Using Docker

Using Docker, all required dependencies are installed automatically. All process tasks are deployed as web services in separate Docker containers and can be accessed through the published ports.

docker-compose up -d --build

Start Services Locally

Alternatively, all services can be executed natively on the local machine. All dependencies, libraries and programs that would be installed automatically within the Docker containers are a prerequisite then.

./raspberrypi.sh

Launch Process

Navigate to the process engine and drop the following JSON process model:

[
    {
        "type": "parallel",
        "name": "",
        "branches":
        [
            [
                {
                    "type": "webservice",
                    "name": "scanning position",
                    "method": "POST",
                    "url": "http://localhost:8084/scanpos",
                    "post": "sleep(5 * 1000);"
                },
                {
                    "type": "webservice",
                    "name": "scan scene",
                    "method": "GET",
                    "url": "http://localhost:8081/obj/universalrobot",
                    "pre": "request.expect = 'binary';",
                    "post": "storage.saveBinary('scene', responseBody);"
                }
            ],
            [
                {
                    "type": "webservice",
                    "name": "select object to find",
                    "method": "GET",
                    "url": "http://localhost:8082/example_models/cuboid5000.pcd",
                    "pre": "request.expect = 'binary';",
                    "post": "storage.saveBinary('model', responseBody);"
                }
            ]
        ]
    },
    {
        "type": "webservice",
        "name": "object recognition",
        "method": "POST",
        "url": "http://localhost:8082/bl?angle=90&height=.372",
        "pre": "request.data = new FormData();\nrequest.contentType = false;\nrequest.data.append('scene', storage.loadBinary('scene'), 'scene.obj');\nrequest.data.append('model', storage.loadBinary('model'), 'model.pcd');",
        "post": "storage.saveJSON('recognition', responseBody);"
    },
    {
        "type": "webservice",
        "name": "calculate robot coordinates",
        "method": "POST",
        "url": "http://localhost:8083/coords",
        "pre": "request.data = new FormData();\nrequest.contentType = false;\nrequest.data.append('recognition', this.recognition);",
        "post": "console.log('move clamshell to: ', responseBody);\nstorage.saveJSON('clamshell', responseBody);"
    },
    {
        "type": "webservice",
        "name": "move clamshell",
        "method": "POST",
        "url": "http://localhost:8084/pick",
        "pre": "request.data = new FormData();\nrequest.contentType = false;\nrequest.data.append('coordinates', this.clamshell);",
        "post": "console.log(responseBody);"
    }
]