Featured

Fonthill Abbey

Posted by on Dec 13, 2017 in Design, Featured | 0 comments

Fonthill Abbey

We launched our Fonthill Abbey website on which we document our efforts to bring this amazing building and the story of its owner back to life. We created an interactive Virtual Reality Game that allows you to experience a forgotten piece of history as you become Lord Horatio Nelson or Lady Emma Hamilton, guests of mysterious Lord Beckford. You will have the rare privilege of entering secluded Fonthill Abbey and meet its extravagant owner, but will you make it out alive?

Read More

Comparison of color measurement accuracy of ColorMunki Design and FRU WR-10QC Colorimeter

Posted by on Oct 25, 2017 in Design, Featured, Research | 0 comments

Comparison of color measurement accuracy of ColorMunki Design and FRU WR-10QC Colorimeter

I am working on a colour project and had purchased the WR10 colorimeter to complement my long serving work horse, the X-Rite Color Munki Design. My ColorMunki is already several years old and I was concerned that its accuracy might have declined. When I measured several hundreds of samples, I noticed that both colorimeters gave me considerably different LAB values.

To determine which device was closer to the truth I measured the 48 defined colours of Datacolor’s SpyderCHECKR 48. I calculated the absolute error both devices made. The results of a paired-sample t-test showed that the ColorMunki is producing significantly less measurement errors on L (t(47)=-9.229, p<0.001), L (t(47)=-4.590, p<0.001) and L (t(47)=-4.871, p<0.001). However, both devices measure colours that are significantly different from the target colour of the SpyderCheckr card on all three measurements. Figure 1 shows the means and standard deviation for all measurement errors.

Figure 1: Mean and Standard Deviation of all measurements for both devices.

There does seem to be some structure in the errors that WR-10 is producing. Have a look at the heat map (Figure 2). The data for my little experiment is available at the Open Science Framework (DOI: 10.17605/OSF.IO/UWEFD).

Figure 2: Heat Map of the absolute errors

Although both devices show some significant deviation from the original, it is not far off from what can be expected of devices in this price range. The ColorMunki Design produces significantly better results than the FRU’s WR-10QC.

Read More

Persuasive Robotics Talk At The Emotional Machines Conference

Posted by on Sep 27, 2017 in Featured, Research | 0 comments

Persuasive Robotics Talk At The Emotional Machines Conference

I was invited to give a talk at the Interdisciplinary Conference on Emotional Machines in Stuttgart on September 21st, 2017. My talk focused mainly on the work I did in collaboration with Jürgen Brandstetter (doi: 10.1145/2909824.3020257, doi: 10.1177/0261927X15584682, doi: 10.1109/IROS.2014.6942730). My main argument was that the number of robots in our society will increase dramatically and robots will participate in the formation of our language. Through their influence on our language they will be able to nudge our valence related to certain terms. Moreover, it will only take 10% of us to own a robot for them to dominate the development of our language.

This is also the first time I used a 360 degree camera to record a talk. This technology becomes particularly useful when following the discussion between the speaker and the audience. YouTube’s 360 video feature does not work in all web browser (e.g. it does not work with Safari). Chrome and Firefox should be fine.

Read More

Bloomberg Businessweek interviewed Omics about my nonsense paper

Posted by on Aug 29, 2017 in Featured, Research | 0 comments

Bloomberg Businessweek interviewed Omics about my nonsense paper

Esmé E Deprez and Caroline Chen from Bloomberg Businessweek visited the headquarters of Omics in India to interview its owner Srinubabu Gedela about his company. Omics is widely considered a predatory publisher that publishes papers without rigorous peer review. Confronted with the acceptance of my non-sensical paper he replied that “Bartneck’s paper slipped through because it was submitted so close to the conference’s deadline.” Yeah, right.

Read More

TicTacToe Playing LEGO Mindstorms Robot Using Computer Vision

Posted by on Jul 17, 2017 in Documentation, Featured, LEGO, Project | 3 comments

TicTacToe Playing LEGO Mindstorms Robot Using Computer Vision

You can play TicTacToe with this LEGO Mindstorms EV3 robot. It uses three motors to drop the balls into the right field. It uses a NXTCam to view the board and then calculates the best move using a MiniMax Algorithm. All future moves are explored an rated according to their winning chances. The work is based on the TicTacToe code of Thomas Kaffka. An IR sensor detects your hand when you drop your ball. The robot is using red balls and the human player uses blue balls. The Java code is available over at Github. The building instructions are available for LEGO Digital Designer. I used the MinuteBot baseplate, which is useful for building static Technic/Mindstorms models.

 

LDD does not have all the required pars in its database. You will have to replace 22961 with 27940. You will also need to add a worm wheel 27938. In addition you should use a lamp to provide consistent lighting. I used a USB powered LED circular lamp the can be powered through the USB port of the EV3. I only had to take out the lens in the middle so that the camera fits through the hole. A rubber band holds the light in place. To calibrate the robot I added a little arm at the end of the base plate against which the robot arm rotates. The position of the camera can be centered on the board using the wrench and through sliding along the axles.

You can also find information about the robot over at Rebrickable. The inventory there is correct and complete. Except for the base plate of course.

 

Read More

LEGO Compatible Medium Sized Thrust Ball Bearing

Posted by on May 24, 2017 in Design, Featured, LEGO, Technology | 0 comments

LEGO Compatible Medium Sized Thrust Ball Bearing

LEGO’s turn table has considerable friction and rotating a model at an exhibition for a whole day would ruin it. A thrust ball bearing is necessary to decrease the friction. I previously 3D printed a large bearing for my Unikitty. For this year’s exhibition I needed a smaller thrust ball bearing so I designed a new medium sized ball bearing. It includes liftarms to hold a worm wheel which results in a rotation ratio of 1:78. You can download the model from A360 and GrabCAD.

Read More