Inspiration and description
We were motivated to build an efficient way of studying existing infrastructure, especially aircraft components as fuselage and avionics equipment (avionics cupboards), and even things like statues and buildings by creating a 3D wire-frame or CAD model of the structure. With such visual representations of the structures available at one's fingertips, multi-disciplinary (civil and aerospace) engineering students and even amateur designers can get a whole new perspective on the design and engineering projects they're working on and even serve as a platform for future inspiration - all of this using a simple consumer drone! Keeping with the aerospace theme of the hackathon, we wanted to incorporate a novel use of the drone to measure the key dimensions of the studied object and create industry-worthy CAD representations of it.
What it does and how we built it
We use a Parrot AR Drone 2.0 to navigate around a structure and record distance measurements from the onboard IR sensors to help create a CAD model containing a set of numerous points traversed on the structure by the drones' flight path around the object. The drone traverses the length of the whole structure and moves around the structure in layers. For each horizontal slice, the drone uses the IR sensors (added using an m-bed microcontroller) to determine the distance from it to the structure. Using the pre-determined motion (which we have chosen to be a square in this case for stability and a simple proof of concept), and the reading from the sensor, we can determine the coordinate for the point the drone is at.
Challenges we ran into
MACROS ALGORITHM Conversion of the coordinate files outputted from the IR sensor required extensive use of macros on SolidWorks in order to render the model using those data points. Since none of the team members had any experience working with macros on SolidWorks, this aspect of the project proved to be a challenging yet extremely worthwhile affair.
FLIGHT PATH OF THE DRONE Sticking to our vision of autonomous drone maneuvers around the studied object, Aditya and Nikhil tried stitching together individual flight maneuvers like hovering, pausing and yaw around a preset perimeter to circumvent around our object (cardboard box). This proved to be pretty difficult because of erratic wind conditions (speed and direction) in the middle of the night and the drawback of the drone not able to correct/maintain the course executed on a Python shell.
We also had trouble figuring out the optimum horizontal, vertical and turn speed ratios that would allow the drone to
Accomplishments that we're proud of
We were successfully able to construct a wireframe CAD model of the objects we scanned. Despite the coarseness and instability of the drone, we were able to eliminate a large portion of the noise and construct accurate 3D models.
What we learned
We learned the power of macros on SolidWorks to beautifully and quite accurately represent CAD models of the scanned objects!
What's next for
Expand this proof of concept further to include more IR sensors on the drone to improve the resolution of the corners, edges and important features (extrudes, cuts) of the object.