Thursday, 27 October 2016

Relvis' NEW body and soul

video
Today in class Relvis performed his amazing routine that lasted 60 seconds. A while back we had to switch Relvis' body because he wasn't charged properly, and the new and improved Relvis was practically an identical twin. In our programming, we had multiple issues in trying to use the light and ultrasonic sensor, and we ended up cutting it out because the file became to big. Some of the comments we got after our performance was to use less repeats and to sync it better with the music. On our last lesson before our performance, our group was having difficulty in trying to reverse which was working the previous lesson. If Loredana, Julia and I had another lesson, we would contain the routine to a small radius and experiment with the other sensors a bit more. Overall, I think our group performed quite well, except when it crashed in to the corner when our group was presenting to the class *internally recoils. Oh well, at least Relvis put some soul in to his performance! To check out my programming go to this link.

To check out more, click here!

Thursday, 15 September 2016

Relvis?

video
Today in class we disassembled Relvis to attach four wheels. We relized quite soon that adding four wheels wasn't an option because it wouldn't achieve anything, so we resorted to elevating him. The final result was unstable, and it looked fairly dodgy in my opinion. The whole of Relvis' body was collapsing inwards and we couldn't fit him in the box.

Monday, 12 September 2016

DON'T FALL OFF THE TABLE

Today we used the colour sensor to detect the reflected light intensity, and we programmed Relvis to stop just before the edge of the table. The task was fairly easy, and the hardest part of the lesson was figuring out the best position for the colour sensor. After we completed the delegated task we moved on to some party dance moves that Relvis was trying out. At the moment, he goes to the edge of the table and says fantastic, reverses, than does a magnificent spin. He can also do a few moves such as going in the path of a triangle or square. I really hope we can program to go in the pathway of an infinity loop which we struggled to do in the first day of programming.

Thursday, 8 September 2016

Traffic lights

Today Relvis learnt the road rules and stopped at a red light. The aim of the task was to go when the colour green was censored, and to stop when red was detected. We had to adjust Relvis' colour sensor position so it could sensor at a higher height. This task was a bit difficult because we had problems with the touch sensor. Halfway through programming, Relvis stopped moving forward in general. Despite conquering the problem and putting the program in a loop, we still couldn't figure out why it suddenly stopped working. Some other problems we faced was that a warning sign was flashing before it could perform the switch and detect red and green. We checked the port connections and even swapped the cables and weren't successful. After a bit of confusion, the problem only laid in a minor mistake of not selecting the colour sensor option. Whoops... In the end we were triumphant and Relvis powered through the intersection course that was set before him.

Monday, 5 September 2016

Colour sensor vs Relvis

In today's class we used the colour sensor to detect certain colours and to behave accordingly. Our group had a few issues with the colour sensor because the censor itself was not angled properly. Also, the port for the touch sensor didn't match what was written on the program, so it wouldn't run at first. The program detected green and yellow easily, and we had no issues with the repeat bracket. For some strange reason, the program wouldn't detect the colour red, and it would either say it is yellow or green. If we had more time to experiment in the lesson, we would of added more colours, but either way we were successful.

Saturday, 27 August 2016

AI technology

Artificial Intelligence (AI) has been a developing type of computer science since the 1950's. The aim of AI is to be able to perform tasks and react to situations similar to what a human would do. Currently, there are no AI that have the complexity and flexibility of a human, but some scientists believe that one day they could be integrated with society, and possibly be our friends.


Herbert the soft drink can robot.
Unlike feature extraction whereby the programmer has to write in what the machine is looking for, AI doesn't need this constant instruction. AI  is currently being used for medical diagnosis, language translation, voice and handwriting recognition and even in schools. At the moment there are simple AI models that are designed to do average tasks. An example of this is Herbert, a mobile robot that roams the office and collects empty soft drink cans and disposes of them. Currently, NASA is using rovers to explore Mars, and AI is even being used in war to clear mine fields. The programmers at South by South West (SXSW) have created an extremely realistic humanoid robot named Sophia, in which they state that they want Sophia to be as conscious and creative as a human. Check out this video to find out more about Sophia.

In the 1950's, a man called Alan Turing worked in the earliest field of AI. He declared that human intelligence can one day be replicated from a robot, and he designed the turing test to prove this. This test involves a human and a robot giving answers to an identical set of random questions, and when the questioner can't identify the difference in structure in the answers, the test was successful. No AI have passed the turing test so far.


These robots can learn in a variety of ways, and a popular method is rote learning. This technique of learning is comprised of trial and error, and the AI applies the past experience to new situations through generalization. Deep learning is a recent method of machine learning. It consists of using a neural network to complete sums and complex equations. Without going into too much detail, neural nets compute answers through inputs, and the system makes itself algorithms. Deep learning was inspired by the human brain, and programmers are further trying to discover how adapt patterns of the human brain to fit AI technology.


To conclude, AI has been developing rapidly over the past 60 years. Maybe one day they would have the capability to react and behave similar to what a human would do, and function in a real world environment.

Okey Dokey

video
Today in class we programmed Relvis to detect obstacles and avoid them. It was fairly easy to program, and it involved putting the steps in a repeating bracket. We completed the first activity in a matter on minutes and decided to add extra qualities such as sounds and facial expressions. After a bit of experimenting with the music that was available, we settled on "okey dokey" which became a catchphrase for Relvis. One of the difficulties we faced with this task is that when we added sound we put in the block before the was supposed to stop. This caused delayed reaction times and it crashed into some obstacles.