Final summary

On March 13^th, we had our final bench. Every group brought their “baby” to give a show. That felling is difficult to describe, with exciting, happiness, hopeful and a little bit sadness. We would say goodbye to our year two project. But before that, we all want to give a brilliant show.

Figure 1: Group members

Figure 2: Year two project bench 

Professor Ralph came to our group first. With short chat about the project we began to deliver the presentation to him. He listened carefully with nodding sometimes. After the presentation we did the test which was quite successfully. Just before the end of the morning part of the bench, our assessor Dr. Ali Ataby came to our group to assessor our project. He listened the presentation carefully and sometimes with funny face. When he heard about our future study might involve Kalman Filter, he looked so surprised.

Figure 3: Dr. Ali Ataby 

Generally speaking, our project is very successful. We realize basic colour tracking, we learned a lot of new knowledge especially image processing related, we broad our horizon as well as enhance our confidence to be a good engineer in the future. “Always begin with easy problem, always have plan B.” will be remembered in our minds forever. 

The following video shows all the function of our Stalker, straight following and horizontal tracking.

Video 1: The test for every function

Summary of Week 5

That was the last lab of the project so we caught the final time to improve the project. We changed the running time and tested the car several times.

Video 1: Test 1-1

Video 2: Test 1-2

Video 3: Test 2-1

Video 4: Test 2-2

Video 5: Test 3-1

Video 6: Test 3-2

In the afternoon, we talked about the poster. The poster was separated into four parts. The poster was designed by Fan Wang, who is a very fantastic photo shop user. The poster was great. We would practice the presentation in the next week.

Week 5 tutorial

On 6^th March we have a meeting with our tutor Pro. Ralph, we were still struggling with the quick reaction of the system. Professor Ralph suggested that we could try to separate the process and set a detection area instead of keeping the item in the middle of the screen.

He also mentioned that we would take the bench in week 6. He would like to see everyone talking about a part of the project.

Week 4 summary

Week 4 summary

On 27^th February, we had the forth lab time. We still found the car turned too quickly.
Professor Ralph popped into our lab during the lab time and had a look of tracking process.

Figure 1. Group member discuss with tutor

Figure 2. Group photo with our tutor

 

In order to solve this problem, we came up with several ideas and had a go. Firstly, we tried to slow down the speed by increase the wheelbase. We got a huge help from the lab teachers- they provided one golden colour metal stick. Then our group member polished the stick to fit with the radius of the neck of the wheel.

Figure 3. Polish the stick

Figure 4: Renewed wheels

 

Because in a certain time the motor rotational speed is constant, the length swept by the wheel is also constant in the same time period. The ratate length equals to radius multiply the angular velocity. So that the longer the radius is, the slower the angular velocity would be in a certain time period. However we found the effect was not obvious. So we abandond the new wheels.

Then we tried to set up a pause time. By setting a pause time, the program will run only in a short time. We can shorten the radio swept by the camera by shorten the rotate time. This idea was better than the one before. In this way, the tracking system will detect the item intermittently but more stable.

Figure 5: The modified programming language

 

 

Week 4 tutorial

Week 4 tutorial

On 26^th February, we had the forth tutorial with professor Ralph, we mainly talked about the problem that the car react too fast with the change of item. Professor Ralph suggested that we change the measure to wheel measure. This will reduce the measure error.

Secondly, professor Ralph said that this project is an open-minded project, it was different from the lab experiment we did in the past. So it was quite different and a challenge for us.

Summary of Week 3

On 20th February, we focused on changing the sturcture of the wheels. In the past test, we found that the wheels of the car could only rotate one wheel at one time. However it would be better to roate two wheels at the same time so that the car could rotate with an axis centre. The main reason for modifing the structure is that in the improved way a more accurate detection could be realised. Here is a video show the car rotate with an axis centre.

Secondly, we tried to change the BRG standard into HSV standard. However we found there wasn't so mand differences in enhancing the accuracy after we changed the form. So we decided to remain the old form.

Moreover, we focused on tracking single item under multi-item background. The programming was not finished. Figure 1 shows part of the Matlab programming language on tracking one dinamic red ball with 4 other stable red ball in the screen.

Figure 1: Tracking one dinamic red ball with other stable balls in the screen

There are some handsome boys working on there section.

Week 3 tutorial

On 19th February  2015, we had the third tutorial meeting with Pro. Ralph. We summarised the prosess of the program and discussed the next step of the project.

Pro. Ralph broadened our horizon. He talked about his understanding of the "track". The continuous changing item is easy to be tracked while the sudden change item is difficult to be tracked. At the same time, Pro. Ralph pointed out that the advanced system was able to predict the path of the item.

The tutorial helped us a lot by providing a new direction to improve the system.

summary of week2

The project had a profound improvement in week 2--the car can be controlled by Matlab to track a red item. We were so excited!
In deatail,
1. Matlab programming: can realise the basic tracking process such as track one red item
2. Arduino: can be controlled by the Matlab well
3. Car :cam turn itself under the instruction of Arduino output
4. The blog was updated

Figure 1 shows the Matlab programming language of camera control.

Figure 1: Parts of Matlab programming lange about camera control



Figure 2 shows the Matlab programming language of red item detecting.

Figure 2: Matlab programming language of red item detecting

The problem we met:
1. The roate speed of the car controlled by Arduino should be modified and improved.
2. The car cannot roate with an axis centre.
3. The system should realise the trajectory prediction

The action for next week:
1. Improve the Matlab to control the car rotate with and axis centre.
2. Change the structure of connection between the Arduino and wheels.

Week 2 tutorial

Today we had the second tutorial of the project.

Pro. Ralph talked about the importance about rules assignment in team working, and hoped that we could finish the allocation as clearly and quickly as possible. Then we talked about the problem we met right now, the camera still could not be detected by Matlab, which was the very basic step of the system. Unless we solve this problem, can the project move on. Pro. Ralph suggested us to work out how to analyze the video file first. After the problem is solved, we will be able to work on the in-time video information analysis.

At the very beginning, what we supposed was to tracking the specific items using camera and Arduino. As the time went by, we found that the project was not that simple. We rearranged the target to detect the red item, cause "specific item" was a huge range, if we could find the way to detect a red item, could we get the way to track items with specific characteristics. Then we limited it to detect the red item in horizontal direction, because we found two dimension tracking needed at least two control structures, which was quite complex. After this, we found there could be several red items in the video which we could not handle with, so we limited it to detect and track one red item with the whiteboard background and lastly we limited it to track one red item with whiteboard in a video file.It seems that we gave up many times, however the real reason for these actions was obey one good idea: we should build the project step by step, little by little. Maybe now we could not solve the complex problem, we are still working on it! We will set all what we abandoned temporarily as our next target and work for it. Hope we can finish the project successfully.

The plan for Week 2

We had a meeting on 10th February night and talked about the plan of week 2.
1. Hao Zhao still concentrates on the programming of matlab.
2. Ziyu Wang and Xinpeng Wang will write the alternative matlab programming language.
3. Fan Wang and Chuntao will work on the information connection of Matlab output and Arduino input.
4. Ziyu Wang will contact with Pro. Ralph while Hao Zhao will contact with Elias Griffith.

A summary of Week 1

On 5^th February, we have a meeting with our personal tutor Pro. Ralph. We talked about the exams first and Pro. Ralph asked especially the common on the difficulty of the exams. After that, we raised a question about our project. We could not find the way to connect the camera with computer. Pro. Ralph provided us a solution by introducing a teacher named Elias Griffith to us.

On Friday, our group member met in the computer lab. We collected the components from the lab teacher and checked them. Unfortunately, we found the Arduino we got was not the type we ordered. As a result, we reapplied an Arduino from the lab and buy another one in the city center. Hao Zhao and Ziyu Wang concentrated on the programming language of Matlab. Hao Zhao wrote the basic language about finding the center of an object. Ziyu worked on the studying of multiObjectTracking function. 

Figure: Hao Zhao

Fan Wang and Chuntao worked on the building of Arduino control board circuit connection.

Figure: Fan Wang and Chuntao

Xinpeng handled on the video importeren process.

Figure: Xinpeng

What we have done in week one:

1. Finished the Matlab language programming of finding the center of an object.

2. Finished the Matlab language programming of analyzing the dynamic frames.

3. Realized the Arduino control circuit building.

Figure: Built up circuit

4. Wrote the Arduino language of motor control.

The barrier:

1. Still cannot find the way to import the information from external camera to Matlab.

2. Cannot contact with Elias Griffith. 

Before the second semester

Our gourp was set up in semester one. The membership included Chuntao Mou,  Fan Wang, Hao Zhao, Xinpeng Wang and Ziyu Wang. The title for our project was "Fire Eye-camera tracking system". The basic achievement about this projest is to tracking an object in the video applying the knowledge of Matlab and Arduino. The design process contains four parts mainly. Firstly, the background knowledge of camera tracking system. Secondly, the code design of Matlab. Thirdly, the code writing of Arduino. Lastly, the combination and checking process.

Professor Jason Ralph is our supervisor thanks a lot for his help. We have met professor several times in December, obtained many good advices.

As the title indicates, a camera is required. Besides this, one Arduino control board, an image collector, a small electric car and Matlab software are required too.

Everything is ready, just look forward to the start of second semester.