Robot cv software

Computer Vision-based Robot Supervision

Disclaimer: Information that is presented here, might be deprecated.

Feature owner: Geri
Date of assessment: 2016-05-08

Current status

• Detects if cargo is present on the railroad and/or in the cargo bay

• Recognizes the markers on the robot using a camera, which is mounted to the camera scaffold and positioned over the robot

• Detects unexpected activity (a.k.a. interference - e.g. hand in the area) and sends an MQTT signal to stop the train

Issues and TODOs

• Determine the rotation angle of the grabber crane relative to the arm

  • This is not blocked by any of the issues, can be implemented anytime

• Recognizing the rotation angle of the cargo

• Create a nice cargo model for the demonstrator. (The current one looks not so cool)

Development directions and plans

• For the camera for this task is placed above the desk, so that this approach is different from the CV based on stereo infra cameras and provides different possibilities for usage

• The main objective of this component is to create an external observer for the system that automates sending control signals related mainly to the robot arm

• The observation of the robot arm wouldn’t be necessary if it worked reliably. This case only the trains and the cargo should be recognized.

• As it was described in the section for the Robot Arm, Geri will work on the project, but assumingly on different tasks.