Rajan Gupta, Department of Electrical Engineering, IIT Madras
INTRODUCTION This report is focused towards transportation industry; affordable personalized mobile vehicle for physically challenged to be more precise. This is being achieved by making a scaled down version of a two-wheeled platform which can intuitively be ridden by tilting the body in desired direction of travel. Standard literature, Inverted Pendulum, will be studied on experimental basis with above stated focus.
THEORY 1. Transfer Function The closed loop transfer function can be represented as !(s) U(s) = G(s) 1+G(s)C(s)
where C(s) = k d s 2 +k p s +k i , and G(s) = ml q s s 3 + b(I +ml 2 ) q s 2 ! (M +m)mgl q s ! bmgl q
!(s) is the angle of tilt, U(s) is external force, C(s) is a controller and G(s) is a open loop representation.
2. Determination of Tilt Angle ! = cos !1 ( bsin" a 2 +b 2 !2abcos" ) where ! is the tilt angle from vertical, ! is the mount angle of sensor from surface, a is the height of the sensor from ground when system is vertical and b is distance of the sensor from ground in the direction of travel of rays.
EXPERIMENTS 1. Sensor Analog distance sensor was used to detect the tilt by measuring aligned distance of a fixed point on the setup from the ground. Various filters like average, mode and median were tried with different filter-tap sizes to remove the noise.
Figure 1: 10-tap median filter
It was found out that median filter, which finds the median of data samples in real time, with 10-tap size was better off as compared to other filters in terms of noise removal, time delay and computational resources required.
2. Maximum Angle of Tilt Impulsive external forces were given to determine the maximum angle of tilt beyond which system will not be able to stabilize about desired equilibrium. As a result of these experiments, it was found to be approximately 12 about mean position.
3. Position Drift It is a translational drift that incurs while system is trying to balance about its equilibrium position. It was calculated by integrating RPM, assumed to be proportional to the PWM output of the PID controller, for a period of 25 seconds.
Figure 2: Position drift as in one of the experiments
The drift in experiments conducted, on an average, was 24cms.
4. Payload Experimental setup was tested for different payloads below 2.45kg , placed at different heights.
Table 1: Range of payload for different heights k p k i k d
0.68 2.56 0.08 500 2450 38 0.765 2.88 0.09 750 >2450 38 0.85 3.2 0.1 1200 >2450 38 0.935 3.52 0.11 1450 >2450 38 Its interesting to note that the setup is stable for a wide range of payload for specific set of PID constants.
5. Implementation Processor: Arduino Mega2560, an open-source electronic prototyping platform, with Atmega128 16Mhz onboard Motors: 200 RPM, 12V DC, Maximum Torque 32kgcm, No-load current 0.8A, Metal Gearbox and Weight of 180gms Sensor: Analog distance sensor with range of 3.1V at 3cm - 0.3V at 40cm Motor Driver: Upto 20A inrush current delivery, 6V-18V operation and simple connectivity to IO pins of MCU Battery: Lithium Ion rechargeable 12.6V 4000mAh, provides more current without any major drop in voltage, discharge capacity upto 8A, light weight and small size Hardware: Aluminum structure suitable for strength and affordability. Setup was designed in AutoCAD before manufacturing PCB Design: Arduino shield was designed in PCB layout and CAM program, Eagle
6. PID Controller Tuning
Table 2: Effects of increasing each of the controller parameters Response Rise Time Overshoot Settling Time S-S Error k p
Decrease Increase NT Decrease k i
Decrease Increase Increase Eliminate k d
NT Decrease Decrease NT Where NT means no-definite trend or minor change
Steps for designing a PID controller are: Determine what characteristics of the system needs to be improved Use k p to decrease the rise time Use k d to reduce the overshoot and settling time Use k i to eliminate the steady-state error
CONCLUSIONS Scaled down version of personalized mobile platform was designed, manufactured and analyzed with limited resources available. It balances with tilt error of 5 . Range of payload for different heights were quantified. However, there are limitations. Such technology is suitable only for flat grounds. Project can be extended to robot formations, surviellence body and ball balancing robots.