Design Modifications

Design Modifications

Image 1: Drawing of potential crank rocker driven system, no gear system or motors included in drawing. 

In order to make the wing function in 3 dimensions, we decided to rock the entire mechanism back and fourth as a plane.  A rocking motion combined with the planar flapping motion retains the proper shape of the 3 dimensional wing tip trajectories found in research. After exploring several potential options to create this motion including several types of 3 degree freedom joints and multi-shafted systems, we decided the best option would be to use the previous ground link as a means to rock the wing mechanism. To do this we planned to make a few additions to the original design. First we would add a second wing to make the whole system visually more similar to a bird. Then we would attach the two planar systems together with a bar between the two as seen in Image 1. Initially we designed for a crank-rocker to control the rocking motion by manipulating this bar connecting the systems. In this case, we would utilize a continuous motor to power the rocking of the system.

The points highlighted in yellow within Image 1 are points that require grounding on some kind of cage not included in the drawing. This cage would be sturdy enough to act as a virtual ground link for the planar wing, as well as house the motor responsible for the rocking motion and be able to handle the force induced by this rocking. One major issue we had to resolve was ensuring that the rotation of the two wings remained planar, that we used the right line of action to consider rotation. In order to ensure that this new rocking rotation can be achieved, the center of rotation for the cage connection points on the wings and the center of rotation for the rocker must be along the same axis. We ended up making the hinges located on the wings attach to the cage on either wing into rings that connected to a hole in the cage. A ring through the gr planar ground pivot point allows the piece to rotate as it did with a dowel in it, but easily connects to the cage through a small hole.  Adding this ring also allows for more variability in position of that rotation point if our alignment is not correct. With more rotation at this point on the upper wing, the original (now virtual) planar ground link may become larger. If this distance, from center of rotation of the driving link on the wing to the rotation point on the upper portion of the wing,  becomes too large our entire device may become a non-Grashoff device causing its performance to suffer.  In addition we were able to find a non continuous motor that allowed for back and forth rotation which negated the need fr a crank rocker device entirely.We also decided to exclude a complex gear system between the two wings to allow for single actuator power source, because we wanted flexibility to shift the wings in towards one another and away if dimensions needed to change on other components.  

Image 2: CAD Model of Final Prototype

As we began to model this system in Solidworks, the difficulties arose from designing the cage. There are several essential details of the cage: first it could not interfere with the movement of the wing, it could not block the view of our device to others to extensively, and needed to be stable enough to support the movement and weight of materials as necessary. Our initial design was rectangular, but quickly the merits of arches was scene in their simplicity, elegant look and obvious ability to stay clear of the wings as they moved. Although our initial design was more baren than that seen in image 2, the premise is nearly identical except with less arch supports, base triangular supports, and longitudinal supports attaching the arches together. These elements will prove useful however as we learn more through fabrication.

Motor Selection

We decided to use the 2 Propeller Continuous Rotation Servos and 1 Futaba Standard Servo to actuate our mechanism. We chose these motors because they were relatively lightweight, and we had free access to them through the UT Austin Advanced Mechatronics lab. The continuous rotational servos have a stall torque of approximately .294 Nm, and the standard servo has a stall .431 Nm. We showed our mechanism to our TA, and after observing how little friction and inertia we had to overcome, we qualitatively decided that the motors should have enough torque to drive their portion of the mechanism. 

Propeller Activity Board

Futaba Standard Servo

Parallax Continuous Rotation Servo

Motor Control

We controlled each of our three motors using a Propeller Activity Board. This board contains five sets of three-pin headers that can be connected directly to small servomotors. The 5V voltage regulator that control’s the devices maximum output voltage and current can deliver up to 1.5A with a 6V power supply. To put this in perspective, the Arduino Uno has a D/C current limit of 20mA for each I/O pin.

Additionally, the Propeller Activity Board contains a microcontroller that is capable of 8-core multicore processing. This ability would allow us to control each motor independently of one another, allowing for the possibility of incorporating very fast feedback loops if we were to use a timing method for synchronization or different PID loops.

After reviewing these specifications, we decided to use the Propeller Activity Board because it would not require us to use an external motor driver to power our motors, and it would potentially allow for tight control loops.

We programmed the Propeller Activity Board in C, and used the Standard Servo library to control each of our motors (a copy of our code is in the appendix). We specified an input speed for each of our continuous rotation motors on a scale of -100 to 100. 100 corresponded to an a full speed CCW rotation, and -100 corresponded to a full speed CW rotation. On the other hand, we specified different set points for the angular position of our standard servo. Additionally, we specified the max rotational speed the standard servo could move at based on how fast we wanted the mechanism to rock back and forward.

In order to recreate this product, it is very important that the continuous rotation servos are calibrated correctly if using the open loop control described above. Small differences in the friction in the motor, the friction of a mechanism wing, and the rotational inertia of the mechanism wing can make each motor rotate at different speeds even when given the same set point velocity.

If given more time to work on this project, we would have incorporated three potentiometers into our design that would be used to regulate the rotational speed of each motor. The user can uses the output voltage from each potentiometer to change the reference speed for the continuous rotation motors and set the max rotational speed for the standard servo. This would effectively create a form of closed loop control using a proportional controller, where the error is completed visually from the user.