Hybrid Power Autonomous Model Vehicle. Yizhao Zhuang. Chike Uduku. Matt Hinnenkamp. March 9, Department of Electrical and Computer Engineering

Transcription

1 Hybrid Power Autonomous Model Vehicle Yizhao Zhuang Chike Uduku Matt Hinnenkamp March 9, 2009 Department of Electrical and Computer Engineering University of Minnesota Duluth Duluth, MN Approved Date Advisor s Signature

2 TABLE OF CONTENTS PAGE Abstract iii I. Introduction 1 II. Design & Simulation 5 III. Hardware Construction 11 IV. Software Development 14 V. Technical Difficulty 20 VI. Future Improvement 23 VII. Applications 25 VIII. Conclusions 25 IX. References 25 LIST OF TABLE Table I. Elements References 10 LIST OF FIGURES Fig.1. Boost converter 4 Fig.2. Power switching algorithm flow chart 6 Fig.3. Power switching circuit 7 Fig.4. Boosting circuit 8 Fig.5. Output Signal compared to input signal 9 Fig.6. Simplified boosting circuit 9 Fig.7. Physical snapshot of the circuit board 11 Fig.8. Boe-Bot robot 12 Fig.9. Solar panels 13 Fig.10. Voltage output of the circuit 13 Fig.11. Boe-Bot front sensors 14 Fig.12. Transmitter and receiver 15 Fig.13. RC circuit for light sensitive navigation 18 ii

3 Abstract The continued use of fossil fuels is putting the earth at risk of significant climate change. Solar power gradually becomes a preferred energy source for many applications. In this project, we applied hybrid power algorithm on the autonomous vehicle combining solar energy and battery energy. The entire hybrid power system is controlled by a power-switching circuit. On the other hand, the micro-processor for the model vehicle is powered by separate battery sets. iii

4 1. Introduction As the population of human beings increases rapidly, an energy crisis emerges. In other words, the amount of energy available per person is getting smaller. To combat this, alternative energy sources are being looked at to supplement or replace current energy options. Solar energy is among those sources that are being looked at as a solution for the future. As the technology improves, solar energy gradually becomes a reliable energy resource for many applications. This project looks to use this technology to supplement the power source of a robotic vehicle. The other half of this project involves the development of autonomy in the robotic vehicle. A robotic vehicle that is able to avoid obstacles and reach a destination without human input could be a valuable technology with many useful applications. Already, such vehicles are being looked at for uses ranging from war applications to unmanned space vehicles. This project looks to successfully enable the robotic vehicle to autonomously travel. History of Solar Energy A solar cell or photovoltaic cell (PV cell or solar cell) refers to variety of electronics which can capture sun light and converts solar energy into electricity by photovoltaic effects. Solar cells have been widely used in many applications such as powering small electronic devices. PV arrays, an array of PV cells, can generate a form of renewable electricity, especially useful in situations where land power system grid is not available while solar power is sufficient, such as earth-orbiting satellites, space stations and deserts. 1

5 Solar technology development can be traced back to 7 th century B.C. when magnifying glass was used to concentrate sun rays to burn ants. However, the world s first solar collector was actually built in 1767 by Swiss scientist Horace de Saussure. In 1839, French scientist Edmond Becquerel discovered the photovoltaic effect while experimenting with an electrolytic cell made up of two metal electrodes placed in an electricity-conducting solution, electricity-generation increased when exposed to light. In 1883, Charles Fritts built the first solar cell. He coated the semiconductor selenium with an extremely thin layer of gold to form the junctions. The device was only around 1% efficiency. Russell Ohl patented the modern solar cell in 1946 (U.S. Patent 2,402,662, "Light sensitive device") [1]. In 1954, Photovoltaic technology was born in the United States developed by Daryl Chapin, Calvin Fuller, and Gerald Pearson using the silicon photovoltaic (PV) cell at Bell Labs. The first solar cell was capable of converting enough solar energy to run lab equipments. Bell Telephone Laboratories produced a silicon solar cell with 4% efficiency and later achieved 11% efficiency. In 1958, the Vanguard I space satellite used a less than one watt PV array to power up radios. Later that year, Explorer III, Vanguard II, and Sputnik-3 were launched with PV-powered systems on board. In 1970 the first highly effective GaAs Heterostructure solar cells were created by Zhores Alferov and his team in the USSR. In 1994, the National Renewable Energy Laboratory develops a solar cell made from gallium indium phosphide and gallium arsenide that becomes the first one to exceed 30% conversion efficiency. In 2006 Spectrolab's cells achieved 40.7% efficiency in lab testing. In the near future, photovoltaic researches and developments will be expected to continue and focused on new materials, cell designs, and novel approaches to solar material and product development. It is 2

6 a future expectation that the clothes you wear and transportation can produce enough power that is clean and safe. Solar power will be used to electrolyze water, producing hydrogen for fuel cells for transportation and buildings. The price of photovoltaic power will be expected competitive with traditional sources of electricity within 10 years [2]. Eventually, solar electricity will be sufficient to meet the required total energy human being needs. Structure of Solar Cell Photovoltaic cells are made of semiconductors such as silicon, germanium or gallium arsenide. We recognize these semiconductors as p-type (positive type), n-type (negative type) and I-type (intrinsic type, undoped semiconductor). The most common solar cell is configured as a large area silicon p-n junction that has one layer of n-type silicon diffused directly with another layer of p-type silicon. Once these two types of silicon are placed contact with each other, a diffusion of electrons will occur from the n-type side with high electron concentration to the p-type side with lower electron concentration. The electrons will combine with the holes on the p-type side once electron diffusing occurs across the p-n junction. The electric field established across the p-n junction will form a diode which promotes current to flow in one direction only. Boost Converter Circuit A boost converter is a circuit which performs as a simple transformer without magnetic flux mechanisms. As Fig.1 shows on the next page, turning on the transistor in the bottom position applies the input voltage across the inductor such that V L equals V in. Meanwhile, i L linearly 3

7 ramps up, increasing the energy in the inductor. This step is called dumping energy into inductors. However, turning off the transistor forces the inductor current to flow through the diode and some of the inductively stored energy is transferred to the output stage that consists of the capacitor and the output load across it [3]. The element CR1 in Fig.1 is a diode. It is used to prevent current flow from the capacitor CI back to the inductor L1. Without the diode, two directions current flowing into each other would cause the current cancellation and energy loss. Fig.1. Boost converter 4

8 2. Design and Simulation The design specification of the vehicle has to be met as follow: 6V maximum for the servos; 110 ma starting current for the servos; 50 ma current in order for the servos to constantly move with normal load. Switching Algorithm In order to give the vehicle hybrid power functionality, we applied a switching circuit. Light intensity value obtained from the light sensor is an important parameter used in this switching algorithm. As the Fig.2 shown below, there are two conditions: the actual light intensity value is greater than (or equal to) the threshold light intensity value; or the actual light intensity value is smaller than the threshold light intensity value. The power switching circuit will switch the energy source to solar panel if the light intensity value is above the threshold value. Otherwise, the battery is used. 5

9 Fig.2. Power switching algorithm flow chart To obtain the value of the threshold light intensity, we used the programmed light sensor to record light intensity values of different environments, such as fluorescent light bulb lighting condition, incandescent light bulb lighting condition, sun light condition, etc. We then recorded the corresponding output voltage of solar panel for each lighting condition. Finally, we determined the corresponding light intensity value by using the voltage value which meets the minimal requirement of powering the vehicle. Fig.3 shows the simplified power switching circuit. One light intensity sensor gives a signal to the micro-controller, the program will perform the comparison as Fig.2 showed. Once the micro-controller finds out whether the light intensity level is above threshold level, it will output a high/low signal to the BJT transistor in order to turn on one part of the circuit. Form Fig.3, there is a NOT logic gate between one of the transistor to the micro-controller. The 6

10 reason of this design is to let only one part of the circuit be on at one time. For example, if the micro-controller sends a high signal to the transistors, the transistor on the left side in Fig.3 will be on, but the transistor on the right side will be off. Fig.3. Power switching circuit Boosting Circuit The boosting circuit is shown in Fig.4. The design of the boosting circuit uses the same algorithm as boost converter circuit does. As we measured, the V-I relationship for solar panel is not linear and the internal resistance of solar panel changes its value due to different load resistances. In this case, we treated the solar panel as a current source I2 in PSpice software. R9 is the corresponding resistance to transform current source into voltage source in the circuit. The element C10 is a 2F ultra capacitor (electrical double-layer capacitor). It is used to keep the signal stable and smooth the oscillations; also, it will store as much energy as it can 7

11 before the energy is transferred to the inductor L1. L1 and L4 are the inductors we manually winded. Each of them has 3 mh inductance and 1.5Ω internal resistance. V1 is the signal source from the microcontroller. It gives a 1 KHz square wave with 75% duty cycle. This signal is connected to the base of IRF 150 (power MOSFET) so that these two different parts of the circuit will be turned on and off in very high frequency. The reason of using a power MOSFET was because the solar panel rating current is 1A, which means there will be lots of heat released at the MOSFET when the circuit is functioning. On the right side of the circuit, C1 is the capacitor to lower the frequency of the output signal. R4 and R10 are simulation of the servos. Fig.4. Boosting circuit The graph Fig.5 shows the circuit output. It takes around 2V input (thinner lower line) and gives almost 5.5V output (thicker higher line) at 4.5 s. The output approaches 5.5V as time 8

12 line moves. Voltage Gain Fig.5. Output signal compared to input signal Fig.6. Simplified boosting circuit As described in Fig.6, this boosting circuit can be simplified when the power MOSFET is off. However, for the case when power MOSFET is on, there is no output signal so we do not need to consider that case. This circuit is linear time-invariant (LTI) system because all the 9

13 initial conditions for inductors and capacitors are 0. Elements in Fig.6 R1 C1 L1 R2 C2 Elements in Fig.4 R9 C10 L1//L4 (L1 parallel with L4 and ignore the internal resistance R6 and R12) R4//R10 C1 Table I. Elements references To find the gain of the system, we set up the equation between two sides of the circuit. V R V V V V in 1 out in out in in out out + C1 = ( + * C 2 ) = dt R2 dt ( s) ( s) dv s + = s + ( s) = 1 + s ( s) R R * C 1 * C * L V * C 1 dv + sl R 1 1 V V out L 1 in dt 10

14 3. Hardware Constructions Upon completion of the design and working simulation portions of the project, we were ready to go ahead with the physical implementation of the circuit. The first task was to find the inductors needed for the circuit. The inductors provided by the lab had too high of a resistance, requiring us to wrap our own low resistance inductors. We then did a preliminary hook-up of the circuit on a breadboard; the image of which is shown below in Fig.7. Fig.7. Physical snapshot of the circuit board Next, we needed to find a low weight option for mounting the circuit on the Boe-Bot (Fig.8). The weight of breadboard we were hooking the circuit up to was too heavy to mount on the Boe-Bot. We decided then to go with a small soldering board. It is light-weight as well as able to fit in an easily mountable area on the Boe-Bot. The circuit was soldered together on the board and was able to be mounted on the Boe-Bot. 11

15 Fig.8. Boe-Bot robot The other part of the physical implementation was getting the solar panels mounted and operational. Each of the solar panels we worked with had rated output current of 1A and output voltage of 0.5V. We put all four of the solar panels in series and soldered their wires together to generate 2V and 1A output. For the mounting of the solar panels, we decided to place the four panels on a piece of cardboard that would be placed on the top of the Boe-Bot, as shown in Fig.9. Cardboard was chosen because it is fairly light-weight and readily available. 12

16 Fig.9. Solar panels The result of the physical implementation was not what we had hoped for. The graph of the input and output of the boost circuit can be viewed in Fig.10. Fig.10. Voltage output of the circuit As can be seen from the graph, our output of 3.28V is well short of the 5V we were able to achieve in simulation. 5V was the minimum requirement to drive the two servo motors. 13

17 4. Software Development The software aspect of this project concentrated on achieving two kinds of autonomous roaming. In the first kind of roaming we set out to make the Boe-Bot be able to detect obstacles and make decisions on how to effectively maneuver around these obstacles. In the second kind of roaming, the Boe-Bot was programmed to autonomously detect when a flash light was turned on as well as navigate based on the movements of the flash light. All components used in achieving the objectives above were programmed using a language called P-Basic and a software called Basic Stamp Editor. The P-Basic language was developed by Parallax Inc. specifically for their basic stamp microcontrollers. The aim of this report is not to shed light on how to program in P-basic, rather to expand on the various concepts which when put together make our objectives possible. The components required for the obstacle navigation were 2 infrared light emitting diodes (LEDs), 2 infrared detectors, 2 servo motors, an RF transmitter and an RF receiver. The components required for the light sensitive navigation were 2 light detecting resistors, 2 servo motors and µF capacitors. Fig.11. Boe-Bot front sensors 14

18 Obstacle Detection Roaming The obstacle detection roaming in this project was designed to be triggered wirelessly. As a result, we made use of a 433MHz RF transmitter and receiver shown to the left and right respectively in Fig.12. Fig.12. Transmitter and receiver Some of the specifications for the transmitter and receiver are: Power: 5V +/- 10% Current: ~10mA normal operation Data Rate: k baud Transmission: 500 ft (or more based on environmental conditions) We had to synch the transmitter and receiver for two reasons. First, the transmitter was programmed on a Basic Stamp (BS) 2px microcontroller while the receiver was programmed on a BS2 microcontroller. The BS2px microcontroller has a faster baud rate (about 972k 15

19 115.2k) than the BS2 ( k). The disparity in baud rate meant that the transmitter and receiver could not communicate properly, i.e, what we put on the data path of the transmitter was not what was received by the receiver. Writing a piece of code that gave an adequate compensation for a mismatch in baud rates based on what microcontroller was being used solved this problem. Secondly, being that this was a serial communication, the transmitter and receiver occasionally went out of phase. Figuring out the adequate time duration to periodically refresh the transmitter by sending pulses to it eliminated this problem. The end result was that we put a sequence of 100 on the data path of the transmitter. The Boe-Bot does not commence obstacle navigation until the receiver recognizes this sequence. The ability of the Boe-Bot to detect and navigate obstacles was made possible by the infrared LEDs and infrared sensors on both sides of the Boe-Bot. The LEDs emit infrared at a frequency of 38.5 khz. This frequency was chosen because the infrared sensors have an electric filter which does not permit light below this frequency to pass through. The result is that interference from everyday lighting sources like sunlight and in-house lighting are avoided. When the light emitted by the LEDs hits an obstacle, it is reflected back and is sensed by the sensors which in turn send a signal to the microcontroller. The sensors send a high state (1) when they do not detect any obstacles and a low state (0) when obstacles are detected. 16

20 Based on the signals received from the sensors, the microcontroller initiates movement in a certain direction by sending pulses for a specific duration to the left and right motor. A pulse duration of 750 sent to a motor keeps that motor stationary. Any pulse duration above 750 (850 being the maximum) rotates the motor counterclockwise while any pulse duration below 750 (650 being the minimum) rotates the motor clockwise. Light sensitive roaming To achieve light sensitive roaming, we take advantage of two things; our knowledge of the behavior of capacitors and our knowledge of the inverse relationship between the light density resistor (LDR) and light intensity. Fig.13. RC circuit for light sensitive navigation In the circuit above, the capacitor is connected in parallel with the LDR. If we charge the capacitor with 5V via the I/O pins (3 or 6) for a short duration of time and then listen, the capacitor will start to discharge through the LDR. The amount of time it takes for the voltage that pin 6 or 3 senses to drop below the threshold value of 1.4V depends on how strongly the LDR resists the flow of current supplied by the capacitor. In essence, if we know the 17

21 discharge time of the capacitor, we have an idea of the strength of the resistance from the LDR. If we have an idea of the strength of the resistance from the LDR, then we have an idea of the light intensity. Based on the perceived light intensity, we can initiate some sort of navigation by sending pulse durations to the servos. An important issue to point out is that in keeping with our theme of autonomy, the Boe-Bot had to know the difference between normal in-house lighting and when the flash light was turned on so it could initiate navigation. This is achieved by taking light intensity readings for both scenarios. An average of six values was used for both scenarios in order to improve accuracy. The result is that when the Boe-Bot senses the flash light turned on, it moves towards the flashlight and its motion can be controlled by changing the direction of the flashlight. If we move the flashlight away from the Boe-Bot, then the Boe-Bot seizes motion and spins in a circle, searching for the flashlight. A summary of the I/O pinots for the various components used for both kinds of roaming is given below: Left servo motor Pin 13 Right servo motor Pin 12 Left Infrared LED Pin 8 Right Infrared LED Pin 2 Left Infrared Detector Pin 9 Right Infrared Detector Pin 0 18

22 Left LDR Pin 6 Right LDR Pin 3 Data pin for RF receiver Pin 11 The code for both kinds of roaming has been attached to this report with adequate comments to aid understanding. 19

23 5. Technical Difficulty As was noted above, the circuit that worked in simulation did not follow correctly into physical implementation. In this section we will examine some possible explanations for this, as well as possible solutions. The first difficulty encountered was reaching the rated output of the solar panels. To reach the rated output of the solar panels required direct sunlight on the panels. In a lab setting this is implausible. We were able to procure high wattage lamps, but these were still no replacement for a sun that doesn t shine often in Duluth, MN during the winter. Second, we faced the problems of modeling both the solar panels and the motors. The solar panels are non-linear current sources, making it difficult to model. The motors were difficult to model due to their inherent inductive properties that were difficult to quantify. Third, due to our inability to accurately model the solar panels and the motors, we were unable to utilize impedance matching within the circuit. With impedance matching, we would be able to maximize the power transfer of the circuit, helping us power the Boe-Bot. Fourth, while we worked with the solar panels, it became apparent that they are not terribly adequate as a source for our purposes. Simply put, the output from the panels would not be enough to reliably, efficiently, and effectively power the Boe-Bot. Finally, our fifth difficulty was with utilizing the boost converter circuit. While we spent countless hours researching the circuit and then experimenting with it, it is possible that we have just not utilized the circuit in the best manner. Possible solutions to these issues can be split into focusing on three different areas: the solar panels, the motors, and the boost circuit. When focusing on the solar panels, we would like to 20

24 be able to get more output from them. We could simply put more solar panels on the Boe-Bot. This would increase either the voltage or current output of the panel array. Along the same lines, we could simply use higher wattage solar panels to increase the output. Another option seems to be implementing a sun-tracking device on our current solar panels to allow it to constantly get direct solar energy. There are problems with these solutions, however. More solar panels or larger solar panels suffer from the lack of available space on the Boe-Bot. There is no more free space on the Boe-Bot to mount additional/larger panels. More efficient panels could be an option, but they are much more expensive and difficult to obtain. The solar tracking device would add another motor to our Boe-Bot, increasing the weight and the amount of power needed to run it. If we focus on motors, we would be looking for more efficiency out of them. With more efficient motors we would need less output from the solar panels, potentially making the current arrangement acceptable. However, more efficient motors would be costly and may not have the power to carry the equipment we have loaded on the Boe-Bot. The final area we may focus on is the boost converter circuit. As we were unable to successfully boost our input to the needed level, we must look to this circuit for answers. One solution could be to simply put the output of the solar panels through another boost stage. If we put the output through a boost circuit exactly the same as the current circuit we have built, the ending output would be 5.6 V. This would be well within our requirements. One problem with this is the additional stage would add too much weight to the Boe-Bot for it to move 21

25 effectively. Also, there is no readily available area to mount this additional stage. The second solution could be to match the input and output impedances. If able to accurately model both impedances and match them, we would be able to maximize the power transfer. We can t know if that would be enough to make a significant impact, but it is something that could be looked into. Finally, another solution could be to simply to research the boost circuit more. As stated previously, even though we put much time into this circuit, we still are not experts with it. It could be beneficial to speak with someone who is an expert with this circuit to get their input and advice. All of these proposed solutions could be areas of possible improvements to the project and could be looked into in the future. 22

26 6. Future Improvement For the hardware side of the project, there are two main areas to be improved for the future. First, using a more advanced solar panel. As stated previously, using a more efficient panel would increase the amount of power output per area of the panel. Hence, we would be able to get more power using the same limited area we have to work with. This would help in getting the appropriate power levels to the motors. The second area for improvement is the use of a control system in the Boost Converter. If, hypothetically, the boost circuit worked as predicted, there may have been instances of inconsistent voltages or of voltages that were too high. To correct such a problem, a control system could be implemented for a feedback circuit to be inputted as the signal that controls the MOSFET transistor. Thus, when the voltage was too high, the control system would lower the duty cycle of the signal, which, in turn, would decrease the boosting capabilities. Similarly, when the voltage was too low, the control system would increase the duty cycle, increasing the boost capabilities. This would do much for stabilizing our output voltage. In terms of the software aspect of this project, an area that could be improved in the future is the purpose of roaming. Right now the Boe-Bot does not roam in any specific direction. However, in the future we could maintain the Boe-Bot s ability to navigate obstacles independently but also make it navigate towards a specific signal. Some ideas from Radar technology can be borrowed to make this happen. With this, we would have a transmitter at a location that is essentially calling out to the Boe-Bot. The Boe-Bot would be in a state of 23

27 continuously searching for this signal until it finds it, at which point it travels to that point. This would help give a purpose to the Boe-Bot s autonomous movement. Another idea is adding a GPS system to improve quality and purpose of navigation. This would again function mainly as giving the Boe-Bot a direction to move in. Instead of following a signal, though, the Boe-Bot would travel to a set of GPS coordinates that would be sent to it. The Boe-Bot would then analyze the coordinates with regards to its own location and travel in the appropriate direction. This, again, would help give a purpose to the Boe-Bot s autonomous movement. 24

28 7. Application An application of the roaming implemented in this project could be in the field of exploration. We could attach a camera to the Boe-Bot and have it wander into areas that are deemed unsafe for human beings. The light sensitive roaming could also have some military applications but would require further development. 8. Conclusion Getting our switching circuit to work properly never fully materialized as we were faced with the constraints of space to mount enough solar panels, total weight of the Boe-Bot and the inability of the mounted solar panels to boost up to the desired 5 Volts. However, as far as roaming was concerned, the Boe-Bot did display a high degree of autonomy in making navigational decisions. 9. Reference Ned Mohan. First Course on Power Electronics. Minneapolis: John Wiley & Sons Solar Cells. Wikipedia. Solar Timeline. United States Department of Energy-Energy Efficiency and Renewable Energy. 25