Project Details

 

 

 

 

 

 

 

 

 

 

 

* This video is incomplete as the team decided to drop the crowd-funding

 

What are you guys upto?

 

We are excited to have you here! Hey- you missed an important point- you. We know that we are bound to fail for sure without you. Our project is about autonomous tracking quadcopter.

 

Why have you chosen this project?

 

Honestly, we need a beginning and the starting point is this.  Also, we strongly believe that this project is one of the best as it gives us an ample opportunity to learn (embedded super-computing, Sensors and networks, machine learning, 3D printing and robotics all converge in one project). We believe this is one of the best ones as the market value is huge (so far, we haven’t come across an imaging based tracking quadcopter (some rare ones are in pre-order stage)). Coupled with this is the fact that we have an ample opportunity to create value around the project. Also, some pricey models (but very innovative ones) are coming up with 'follow-me' option alone- not 'follow anything'.

 

Do you have intentions to compete with others?

 

Honestly, we have no plans to compete with any established project. Our intention is to disrupt any field that requires attention through innovative technologies. In case a project exists already, we either build over it with advanced features or simply switch the domain of project. We have chosen this because even if some are in pre-order stage, our solution be different and we are building on existing harwares like Ardupilot and open-source softwares (yet to be finalised).

 

What impact are you planning to create?

 

Our intended impact level is at three different levels –

 

At first stage the team is going to be impacted directly. We will be able to learn technical stuffs and learn the art of management.

 

At second stage, we reach a bigger level of impact- backers (perhaps like you) by empowering them to be better than us by providing resources that we have come across and resources we have created so that they do not need to re-invent the wheel. We will be able to empower these people to impact better than us. Also, we have hopes that these people will be able to spread the awareness and thus our project. We are planning to hold a workshop and sponsor one year’s education to 2 really needy PU students who are really interested in programming as a whole (We have contacted several charity foundations and hoping for a response).

 

At third stage, we reach out to the general masses- we are planning to make our softwares, 3D prints and PCB layouts (if time permits us) completely open-source. This is direct impact. But, indirect impact in potential deployment as outlined in the presentation looks much more promising.

 

You suggested that I was the most important part. Where am I in the picture?

 

Oh sorry! You are absolutely needed for this. There are several ways to join in the revolution. One way is to avail the discounted resources (perks). Other way is to spread the word about the awareness that you may be really wanting to.

 

Note that, you will not find a quadcopter in the perks. We believe not in providing food directly but in teaching people to raise crops so as to feed themselves with self-esteem. Permanent solutions are what this team is counting on. The best part is that you will be provided with materials so that you can be far better than us.

 

Have you assessed the market value?

 

Yes. Details of our assessment can be found in the presentation. (We recommend viewing presentation in full-screen)

 

What are the technological innovations that you are attempting?

 

We are planning to incorporate embedded supercomputing, 3D printing, machine learning and networking all-in-one project TraQuad.

 

These are tentative currently as many of newer practical concepts and the practical feasibilities arise only during the implementation. (We are hoping that our plans extend with better algorithms and better plans along the way during the implementation)

 

3D printing is being used planned for deployment as shielding. Quadcopter rotors are very vulnerable to damages hence, we are planning to make it resistant to bumps by covering the rotors. Plans are on to include some sort of covering for the sensitive electronics.

 

We are viciously cutting down processing costs to include heavy processing on a single board. We are coming up with own algorithms. Two of the team members have developed their own path-planning algorithms (as a part of TI-IDC 2015) a part of which is being planned for use in this project. [Path planning algorithm has been able to obtain shortest path for a 2.5km x 2.5km area in 100 milliseconds in MATLAB]

 

Regarding image processing, we are planning to incorporate two cameras – one in the outright bottom part of the quadcopter and the other one in the ‘front’. To measure the depth, we are deciding between low-cost analog proximity sensors and ultrasonics. [Ultrasonics are generally vulnerable to the noise generated by the rotors themselves]

 

In ‘image processing’ we are planning to discard ‘image processing’ as it doesn’t cater to the real-time processing needs. Instead, we are planning on ‘video processing without image processing’. Currently, we have implemented averaging filter (linear filter) and median filter (non-linear filter). We are also exploring through correlation between images directly so that redundancy in processing images individually is eliminated.

 

We are working on OpenCV platform as it is almost 50 times faster than MATLAB. But, however, there appears discrepancies between older cv library and newer cv2 library which are to be configured for cv2.

 

We have designed a 3D model of the quadcopter to check out the feasibility of the designs. However, we are almost certain and sure that the cost can come down to 10$ or lesser as we are planning for a much better design involving minimal components.

 

We are planning to use machine learning in three places namely ESC (fancy acronym for Electronic Stabilisation Control) so that flight stabilisation is good, stabilising image (we are assuming the PID response to be little unfavourable (practically worst case)) and in prediction of the position (along with kalman filter or extended kalman filter)

 

We are developing app on MIT app inventor (as development is really fast) but, the constraints posed by this is forcing us to resort to Android Studio which is looking really good.

 

We are planning on on-board processing alone once the box is just clicked.

 

We are planning to chuck gimbal and instead use switching mechanism between two cameras.

 

What is your budget?

 

We are estimating it at 400$ currently. The hardwares are costing us 350$ already and we are about to finalise it.

 

Do you have traction?

 

Currently, we do not have any traction and this is the starting point – we are excited despite little traction because of the impact that we are about to create if this campaign goes successful (which we think will surely happen).