Week of 11/17/14 Progress Report

      This week was focused, again, on research. Since choosing our drone we have discovered that we need to learn more about its operating system and how we are going to be able to program it, so we started learning about ROS (Robot Operating System) using this document: http://www.cs.utexas.edu/~todd/cs378/slides/Week8a.pdf.
  We also took it upon ourselves to learn more about the drone that we chose and its components. We used this website to accomplish this task: http://linuxgizmos.com/linux-powered-quadrocoptor-has-three-cameras/
Moreover, we started looking more into wireless charging, because we believe that it can be much more efficient in out project if we charge the drone wirelessly instead of having it dock at a station every few minutes to recharge. The following 2 websites were used to research wireless charging: http://www.wirelesspowerconsortium.com/technology/magnetic-resonance-and-magnetic-induction-making-the-right-choice-for-your-application.html
http://powerbyproxi.com/wireless-charging/
       We also completed the system architecture for our project. It describes how every component of the project is related and will hopefully be useful to us in the future when we really start developing the project. Also, we started developing the system block diagrams in order to explore different paths to accomplish our ultimate goal, which is to have an efficient and easy to use drone delivery system. Those diagrams are not complete, but they will be in the coming days.
        However, our project did encounter some obstacles. First, when we chose the drone, we only considered its use for our project and failed to take into account the price. While it is not over our budget, it does come close to our limit. It is now priced at around $900.00. Our plan to solve this issue is to find parts that we could make ourselves with the 3D printer, such as the protective ribbons that can be ordered with the drone, thus reducing the cost and allowing us to tailor these parts to our own liking/need.
        Our other problem, and perhaps the most important, is the fact that we are behind schedule. By now, we should already be designing our project and starting to develop it, as shown in the gantt chart*. Ideally, in a couple of weeks, we would start designing the algorithm, but at this rate, it might be tough to do that. Our plan is now to break up the tasks that still need to be done into smaller, more manageable portions, and then distribute them between group members so that they can be done more quickly. Hopefully, if our plan works, we should be within our schedule in 3, maybe 4 weeks.

        Link to Progress Report Presentation: https://docs.google.com/a/erhsnyc.net/presentation/d/1MeY274P74Ll9ivr7eyqvbRyb9BDD5HLxfVg7-sqmIxE/edit?usp=sharing

*I know the gantt chart is hard to see, but we ran into some technical difficulties with the software and could not get a better image.

Sensors and More Drone Information

This weeks work was focused on finalizing our decision for the drone we want to use. In the end we decided to take Spiri over the other drones, simply because of its open source platform. We spent additional time looking at and analyzing dimensions, sensors and properties of the drone we selected. We gathered some information from outside sources on the performance and ability of the Spiri. The work done this week was very similar to last weeks, but we focused more on the abilities of the sensors of the Spiri. These are some of the sources for our information:

1. http://spectrum.ieee.org/automaton/robotics/aerial-robots/spiri-programmable-quadrotor-drone-kickstarter
2. http://thechronicleherald.ca/business/1147707-flying-robot-developer-plans-spring-launch-of-spiri
3. http://www.hackthings.com/meet-spiri-your-autonomous-and-social-creature-that-flies/
4. http://www.gizmag.com/spiri-quadrotor-drone/28700/
5. https://www.kickstarter.com/projects/914887915/spiri?ref=nav_search
6. http://pleiades.ca/about/

A very significant part of our research is going to be on ROS (Robot Operating System) because it will provide the basis for our design and development of the algorithm. On separate note, the other time of the week was spent researching the capabilities of the sensors on board.

1. GPS - Can be used for a variety of things when it comes to keeping track of positioning. To be more specific, the drone will be able to be guided anywhere in school with up to 3.5 to 4 meters accuracy. Now this is not a small distance by any chance, but instead for full on coordination with GPS, it will be used partially to set waypoints/end destinations/docks/etc.
2. 9 axis IMU - Essentially can be used for anything force related when it comes to sensors. It probably wont play a major part in the algorithm, but will serve as safety precaution when it comes to emergencies. A page with PLENTY of intimation can be found here. Basically explains the basics of gyroscopes and accelerators. http://www.starlino.com/imu_guide.html
3. Ultrasonic Sensor - In the case of Spiri, this sensor is located on the bottom of the drone so its major purpose will be to detect height and changes in height. The major benefit of this sensor being on the bottom is that it will extremely useful for docking/parking at pre-launch or landing. 
4. Cameras - By far the most important tool the drone contains. The Spiri comes with 3 different cameras that can film at 1080pi at 30fps. These stereoscopic cameras can be used for a variety of different functions when it comes to sensing the surroundings. First this video provided the basis for what exactly "stereoscopic" means when it comes to cameras and lenses. https://www.youtube.com/watch?v=UOnqoC-dJcg Additionally, the cameras would be extremely useful in the field of Computer Vision. Using this technique, we can analyze points on the wall to create a dynamic path to the designation. A video that I watched was https://www.youtube.com/watch?v=715uLCHt4jE. It explains in depth the basics of computer vision. I will continue to watch the later lectures as well. 

In summary, advances were made in the research of sensors and how they can be used for our purposes. More research is going to be conducted on more advanced methods for maneuvering in 3D space. From here we begin our design for the algorithm along with the possible ways of its implementation. 

Selected Drone and Sensor Analysis

This week was heavily focused on researching drones that could be implemented in our project without going off task. Initially, we looked for prices and eliminated the ones that were over our budget, because regardless of their specifications, they were out of reach. Next we analyzed their sensors, flight time, size, work load and if they were programmable through an already established API. Some of the drones were eliminated because of the prices, straight out. However, some did not make the cut because they lacked the necessary sensors and/or platform for programming. The 9 drones included the Parrot Beebop Drone, the DJI, the IRIS, the Storm Drone 4, the Quanum Nova, the HEXO+, the Spreading Wings WooKong-M, the Phenox and the Spiri. Individually, Gui and I focused on researching the Spiri and Phenox. Arianna spent time researching the Quanum Nova, the HEXO+ and the Storm Drone 4. Sebastian spent his research time on the Parrot Bebop Drone and on the DJI. Lastly, Ben did very extensive analysis of the IRIS. In conclusion, we determined the Spiri to be the most suitable for our project, at least at the current state. It provides an open source programmable environment, has an array of sensors and cameras and also seems stable and reliable. The second drone that was very close to getting picked was the IRIS. It was by the far the largest drone with lengths of 21 inches from motor to motor. The IRIS is able to carry nearly 4 times as much material as the SPIRI, but is rather bulky and large. This being the sole reason we did not pick it. The idea here is to start out small. Once we know its doable at a smaller scale, it could easily be implemented with more advanced and heavy weigh drones.

Decided to take this drone was not easy and the decision itself came with a lot of problems and questions. Initially we needed to determine the load the drone was to carry, regardless of what model it is. Carrying important documents or pens is not a big deal, but having to take a stack of papers could be problematic. It was a hard decision to make because the SPIRI provided a wider arrangement of tools for programming, even with its lower load. The IRIS mentioned vague information on its sensory system, things along the line of "automatic avoidance of obstacles" or "returning to start position". This provided no solid information on whether we can design algorithms and upload them. The price consideration was also another problem. We did not want to take more expensive drones  for the sake of their price. If something was cheaper and functioned better for our needs, it would be more valuable.

Starting with next week, we are going begin researching how to use the sensors attached to our selected drone in the most efficient way. This will include ultra sonic range finders, infrared sensors, accelerometers, altitude sensors and other IMU's that are specific to the drone. We will also spend time learning about methods using the camera for a sensor, and how it could be used for positioning in 3D space. We want to make sure that the sensors that are coming with drone are absolutely acceptable for the job we need to do. In general, I think spending time researching the right drone will lead to more efficient developments in the future.