2015 Summer News

CCTV Camera World

Semi-Autonomous Upgrade with Pattern Recognition Camera

For the 2015 Jerry Sanders Design Competition our robotics team was presented with the challenge of manipulating a set of various objects in a controlled field. In order to do so the team re-designed a robotic arm that was up to the challenge.

The team also decided to switch from the previous shoulder mounted control system to something entirely different in the hopes of creating a more intuitive and robust system that would also be more modular. Our software team decided to program a gaming controller in order to manipulate our robotic arm. While this provided us with more flexibility as far as modifying our controller layout versus having a fixed controller, it did not fully address our issue of having intuitive controls and we were left with the question of how we could improve our control system further.

cctv_one.png

2015 Arm Skeleton


Making our competition robot even semi-autonomous had previously been limited by the cost of the technology that is required. This included a camera that had high resolution capabilities and fast data transfer.

But because of our new high resolution 5 megapixel IP cameras from CCTV Camera World, IEEE @ UIC Robotics is now up to the challenge of creating a semi-autonomous arm controller.

The type of cameras we need for the object recognition must be sophisticated, powerful, and have the ability to capture and stream a lot of visual information as quickly as possible.

cctv_two.png

5 Megapixel IP Camera 1

Through the implementation of this semi-autonomous system, the user’s job becomes much easier. All that is required of the user is to drive up to an object and then the autonomous portion of the control system detects the object and picks it up instead of the user having to fine tune the arm positioning in order to pick up the object. A semi-autonomous implementation like this completely addresses the team’s concern with intuitive controls.The cameras obtained from CCTV Camera World are exactly what we needed, offering up to 2592x1944 pixel recording at 10 fps!

Although video output from these cameras is meant to be for a security video recorder, we were able to modify it for our use. Using software from 123 CCTV we were able to convert the proprietary video stream in to a open source RTSP stream accessible by our software.

cctv_three.png

Utilizing such a system consists of the camera sending information about the robot’s surroundings to an onboard computer in order to determine if there are objects of a specific shape within reach of the arm. If there are, the camera then helps gather data required for the arm to automatically pick up the object on its own.

Topic revision: r2 - 2015-06-01 - 18:50:54 - Main.pnowak5
 
Copyright 2016 The Board of Trustees
of the University of Illinois.webmaster@cs.uic.edu
WISEST
Helping Women Faculty Advance
Funded by NSF