We’re hard at work on developing and bringing the world of IoT to your raspberry pi, and we’re excited to show off some of the work we’ve been doing. Today we released a video showing compatibility with an LED matrix screen. While this alone is a cool use of the technology, essentially being able to draw on the matrix to show patterns, there’s something else unseen that is being shown off in the video: Speed.The video shows both the cell phone and the Pi right next to each other, but they are communicating half a world away! The mobile device is sending commands from Poland, to a server located on the West coast of the United States, which is then sending the commands back to the raspberry pi, again in Poland. Once the led status has changed on the pi that information is published back through to the server in the West coast, and back to the cell phone. That’s over 10,000 miles of a round trip in almost as much as it takes you to blink. All this traffic secure and encrypted. IoT One Cloud from Tigase is poised to give you internet connectivity to devices you’ve never thought could be controlled, or deliver information over the web. What will you create?
Many of our users and customers received email notifications from our system about changes to their accounts, such as email change and other information.
After investigation it turns out these email were generated by our development system on which the team decided to change all users’ email to avoid spamming them with unwanted email notifications which could have been generated during testing new features.
Unfortunately this resulted in the exact thing we tried to avoid. Flood of emails notifying our users and customers about changes to their accounts.
We are very sorry for the inconvenience and confusion and unnecessary emails.
The good thing is that there is nothing to worry about, accounts are not hacked or compromised.
Check it out, our second IoT One Video!
IoT to any device you can imagine through a Raspberry Pi!
First of all after creating valid pom.xml and including all dependencies and native librariesI realized that openCV330.so file included in jar file has to be build directly on raspberry pi.As ".so" file has to be build for specified architecture. In this case ARM processor. Process of building openCV dependencies is the same for all devices and is posted on wiki here OpenCV.
Another problem was related to lack of support by java-8 for ARM processor. So it required to redesign code from javaFX to java swing.
And the last issue was caused by lack of v4l2 on raspberry pi. It's really misleading as typical camera test "raspistill -v -o test.jpg" doesn't require v4l2. But every attempt to run camera by java application will fail without v4l2. Fortunately it can be easily fixed with few lines of code and is explained here Camera and V4L2.
In last 2 weeks. I made some minor changes in code added comments and started deploying software to robot. I made connection with RaspberryPi via ssh for quick access to computer and via vnc to observe camera actions. I started learning abut maven and gradle. What took me the most of time was creating a java openCV jar file what isn't finished yet. I'll add code to repository as son as I'll check if it works properly on raspberryPi.
We have changed content format from Textile to Markdown today. Please make sure you use correct formatting markup.
All existing content has been converted automatically but some formatting tags or more complex structures might not be converted or may be incorrectly converted. If you spot any places which needs corrections, please let us know.
Colour detection program is now able to find eg. tennis ball in a room and shows it's position and size. I made some changes to increase performance and now its has max and constant value of 30fps all the time on PC. I also changed a way how program capture centre of ball. Now it tries to find minimal bounding rectangle then counts centre of that rectangle. Instead of trying to use openCV "moments" function. This changes caused that program is more resistant for errors caused by smaller objects with similar colour and can capture more remote objects. There are required only few changes in code.
This week I spent few hours trying to create aruco markers detection in pure java. Unfortunately it's only supported in c++. I decided to postpone it to aim at colour detection instead, to provide reference point for camera vision. I created javaFX GUI reading and displaying image from camera which is able to detects colours and shows edges of given object in real time. To provide easer colour range selection I added six sliders to manually adjust HSV range parameters what's very helpful due to not intuitive HSV colour representation and not constant hue and saturation on different cameras in different light. Next step is to find center of mass of detected object to give exact XY position for steering purposes.
Monsterborg was fully assembled. I made initial investigation on aruco markers detection in java openCV. I created first basic program which runs camera vision, converts openCV "mat" format to java BufferedImage and finally shows video in swing window. I came across a serious problem regarding computer vision in java. As pure java doesn't support aruco markers detection.
Also available in: Atom