The Crane With a Brain.


Group Picture
Members:
IYENGAR, PRAHALLAD
RAMIREZ, MARC
SVENSSON, ELLIOT
THORN, DORIS



problem formulation


problem formulation mode A


problem formulation mode B




Brief Description:

Our project was to design, build, and control a model of a device that would assist workers in a factory. The goal was to build a machine that would allow an operator to move a heavy product or piece of equipment easily without using a push button controller. The machine was designed so that an over-head cart could ride on a one-dimensional track. A cable hangs from the cart so that the mass can be attached. The cart is also attached to a belt that is powered by a motor. The motor has an encoder built on; this is the only sensor for our system. The machine has two different modes of operation. The first mode (Mode A) attempted to magnify the operator's strength. This means that the mass would move when pushed and stop when not pushed, but the amount of force required would be much less since the motor would assist. The second mode (Mode B) mimicked an assembly line. The operator could move the mass from station to station by giving the mass an impulse. Each time the mass received an impulse, the mass would move to the next station. Overall, the machine functioned well and more importantly brought up important questions and facts that would have to be considered in future, follow-up projects.



Control Diagram
Control Diagram


Control Systems:

We set up two different control algorithms for our machine. The first attempted to assist an operator in moving the mass. This algorithm is called mode A. Mode B moved the mass down an "assembly line" when it received an impulse.

Mode A:

Our goal was to assist the operator by driving the angle of the cable to zero. Since we couldn't measure the angle, we assumed it by the angular acceleration of the encoder. We implemented a PI controller that drove the acceleration of the cart to zero. This controller was constantly running.

Mode B:

Mode B made use of a state structure in order to function. There are three states: wait State, move State, and control State. The algorithm began in wait State. In wait State, the DSP simply waits for a measurable change in position that would be caused by an impulse. Once this impulse is felt the code transitions to move State. Move State jogs the motor at a given duty cycle for a pre-determined distance. Move State transitions to control State once it reaches it destination. Control State implements a PID controller on the position of the cart in order to damp out oscillations. The controller runs for a given time and until the cart is within a given tolerance. Once both of these criterions have been met control State transitions back to wait State and the cycle can begin again. The diagram below shows this structure.


Picture of our system
Photo of our system



Important considerations for follow-up work:

Safety considerations
· Should there be a way for the system to sense if someone is in front of the mass while it is moving?
· Should there be a maximum speed of the mass?
· Should there be a sensor to measure the amount of oscillations in the direction we can't control?

Physical Machine
· Should it be larger? For example able to hold 50 kg instead of 5kg.
· Should effort be put in to decrease friction in the track?
· Is there an easy way to cheaply and reliably measure the angle of the cable?

Mode A
· Instead of assuming the acceleration of the cart is only produced by the angle of the cable, the calculations should also take into account the dynamics of the motor.
· Is there a more effective linear controller then our PI controller?
· If not, is there a more effective non-linear controller then our PI controller?

Mode B
Is an impulse the best way to determine that the operator wants to begin the mass moving?



Learn more: [Bill of Materials] [Encoder Datasheet] [Motor Datasheetl] [Wiring Diagram] [Drawings] [Pictures]