Elektroda.com
Elektroda.com
X

[Line Follower] Impact

sailo 18201 22
This content has been translated » The original version can be found here
  • [Line Follower] Impact

    Introducing the Line Follower class robot named Impact. This is an improved version of previous design described in the largest forum of Polish amateur robotics. The robot was created in 2011, until now he was the winner of all competitions in which he participated. The biggest success is undoubtedly the first place in the international robot tournament in Vienna, called by many unofficial European championships. The robot consists of two modules: the main board and the sensor plates, connected to each other by means of light carbon strips. The weight of the whole with a battery is 105 g.

    Sensor module

    This is the element furthest from the center of rotation. The moment of inertia is large (mass multiplied by the square of the distance from the center of rotation), therefore, in order to be able to extend the sensors far, the mass of the plate should be as small as possible. In the previous version, we used 19 sensors, 16 of them in arc, 4 forward and back respectively. The plate also contained comparators and a potentiometer for setting the threshold value (Figure 1).

    [Line Follower] Impact
    Fig. 1 Sensor system in the previous design.

    In the new module, we used 14 sensors, placed at greater distances from each other. As a result, the span of extreme sensors remained unchanged. Sensors moved forward are eliminated. It turned out that at high speeds, taking into account the inertia of the drive, the robot was unable to react effectively to the signal from them. Due to the reduction of the number of sensors, there was the possibility of using the analog-to-digital converter built into the microcontroller (16 multiplexed inputs), which allowed to dispense with the comparators. Without these additional systems, the plate sizes have been reduced. We also changed the thickness of the 1.5 mm laminate by 0.8 mm. These treatments led to a two-fold reduction in the mass of the plate from 8 to 4g. The appearance of the module is shown in Figure 2.

    [Line Follower] Impact
    Fig.2 View of the sensor board in the current, new version of the robot.

    Reflecting transducers have been used to detect lines. KTIR0711S . Connected in groups: three sensors in series with a resistor. Pads for the digital distance sensor have been placed on the board Sharp 40cm < br />.

    Main module

    The board is both a printed circuit and a construction chassis. In addition to electronic circuits, we have included propulsion engines and a tunnel drive. We managed to significantly reduce the module in relation to the previous version of the robot. The dimensions are: 140 mm x 60mm.

    Electronics

    The heart of the robot is a microcontroller from the native STM32. The motors are controlled by H TB6612 bridges. The power feed consists of a 5V impulse converter and a 3.3V linear stabilizer.

    Microcontroller - 32-bit STM32F103RBT6 with the ARM Cortex-M3 core, including among others : 128kB Flash, 20kB RAM, USB, CAN, UART, I2C, SPI, ADC, DAC in the LQFP64 housing, fulfills the following tasks:
    readout of input port states,
    converting analog signal into digital form,
    generating PWM signal,
    control bridges H – generating appropriate signals,
    implementation of the control algorithm,
    communication with the LCD module,
    control of LED diodes. [/ list: u: b4e32f31d6]

    Motor controllers - two two-channel Toshiba H bridges TB6612 , enabling:
    speed control by means of PWM signal,
    changing the direction of rotation of the motor by changing the states of two pins,
    fast braking. [/ list: u: b4e32f31d6]

    To protect against system damage at maximum current consumption by motors (1600mA), bridges A and B were connected (current efficiency increased to 2A). The bridges have been connected as shown in Figure 3.

    [Line Follower] Impact
    Fig. 3 Motor controller connection diagram.

    Remote control

    Due to the high speeds achieved during the journey, it becomes more difficult to stop the robot manually. Starting and stopping is done wirelessly, using infrared. The ATtiny13 system used is responsible for decoding the signal from the remote control, which transmits the signal in the RC5 standard. The solution developed by Philips increases the resistance to interference from the environment and creates the possibility of using universal and publicly available remotes. Another advantage is the ease of using additional buttons on the remote control. An additional processor has been used due to the high requirements for the reliability of remote stopping operations, e.g. in emergency situations.

    Drive

    The drive consists of two engines Pololu HP   with transmission 10: 1 with the following technical parameters:
    Idle speed at 6V: 3000rpm,
    Idle speed (6V): 120mA,
    Peak current: 1600mA,
    Torque: 0.3 kg * cm (29 mNm),
    Dimensions: 24 x 10 x 12 mm,
    Weight: 10g. [/ List: u: b4e32f31d6]

    The wheels consist of rims made of plastic polyamide and specially selected tires. The rim fits tightly on the motor shaft and protected with cyanoacrylate adhesive (Figure 4)

    [Line Follower] Impact
    Fig. 4 Wheels made of polyamide with Mini-Z tires.

    Different types of tires have been tested. The best choice turned out to be the tires used in the Mini-Z car models. These are 12mm wide and 3mm thick tires. The next parameter is the hardness, which is 20 ° (on a scale of 10 ° -60 °). Tires with lower hardness are characterized by greater adhesion, but they wear out more quickly. The diameter of the rim with the tire is 27mm. The cleanliness of the tires is also very important. Before each passing, they are cleaned to remove dust particles that cause loss of adhesion and have a negative effect on performance.

    Theoretical maximum linear speed of the robot in the limit of 3m / s. Depending on the route, the average speed achieved is 2.3-2.5 m / s.


    Tunnel drive

    An important element is the EDF tunnel drive. It is a turbine, as it is used in flying models, but mounted inversely. The element is to create an additional pressure force, which helps the robot to stay on the route in curves at high speeds (above 2m / s). The turbine is equipped with a brushless motor (11,000 rpm, power consumption about 4A), which is controlled by the company's controller Dualsky .

    [Line Follower] Impact
    Fig. 5 Tunnel drive EDF27 with engine control.

    Power supply

    To power the robot the package was used Lithium-Polymer Dualsky 220mAh 25C 7.4V (Figure 6). The continuous current that the package is able to provide is 5.5A, while the peak current is 11A, it is sufficient for proper power supply. The battery allows for approx. 30 seconds of optimal driving, after which the supply voltage drops which negatively affects the dynamics and maximum speed of the robot. During the competition, battery replacement is usually performed every 2 rides, which allows the use of full engine power. The mass, which is about 16 grams, had a big impact on the use of such a small package.

    [Line Follower] Impact
    Fig. 6 Dualsky battery 220mAh used.

    Directly from the battery are powered by motors and a tunnel drive. Electronic components that require 5V voltage are supplied with stabilized voltage using an adjustable converter. ST1S10PHR with current efficiency up to 3A. The processor's power, or 3.3V, comes from a linear LDO (low-dropout) system. LF33CT , whose input voltage is derived from the 5V converter. The power scheme is shown in Figure 7.

    [Line Follower] Impact
    Fig. 7 Block diagram of the power path.

    User interface

    Setting up the regulator requires frequent changes of parameters such as maximum speed of driving motors, turbine rotor speed or PID controller gain. Connecting the robot to the computer after each pass, especially at competitions where the service stations are located at certain distances from the route was very troublesome. A module was created with an LCD display for viewing the settings and buttons for adjusting them. As mentioned earlier, mass is a key parameter, therefore the system is a separate module that connects to the robot via the UART interface.

    The main functions of the module are:
    choice of regulator's settings,
    choosing the maximum speed,
    selection of rotational speed of the tunnel rotor turbine rotor,
    checking the correct operation of the sensors,
    viewing the output data of the PID controller,
    setting the threshold voltage value for the reflective sensors. [/ list: u: b4e32f31d6]

    [Line Follower] Impact
    Fig. 8. Module with LCD display.

    Software

    The software was written in C language using the libraries provided by STM: STM32F10x_StdPeriph_Lib_V3.5.0. The control algorithm is PID with some modifications.

    Achievements so far:

    1st place T-BOT – Wa³brzych - 2012
    1st place Robomaticon – Warsaw - 2012
    1. the place of the Robot Challenge – Vienna - 2012
    1st place of the Tri-City Robot Tournament - Gdańsk - 2012
    1st place CybAirBot - Poznań - 2012

    Photos konstrukcji

    [Line Follower] Impact [Line Follower] Impact [Line Follower] Impact [Line Follower] Impact [Line Follower] Impact

    Filmy z zawodów


    Link



    Link



    Link


    Autorzy konstrukcji:
    Bartosz Derkacz
    Szymon Mońka

    Cool! Ranking DIY
    Can you write similar article? Send message to me and you will get SD card 64GB.
    About Author
    sailo
    Level 9  
    Offline 
    sailo wrote 6 posts with rating 43, helped 0 times. Live in city Wroc³aw. Been with us since 2012 year.
  • #2
    dondu
    Moderator Design Microcontrollers
    So this year belongs to you - Congratulations! :)

    1. How did you choose the distance between the sensors and the wheel axis (center of rotation)?
    Experimentally, or specific calculations?

    2. What is the element connecting the sensors with the motherboard?
    Carbon fiber?
  • #3
    gbd.reg
    Level 21  
    Hello, great design, congratulations on achievements ;) I would like to note that the first two drawings were inversely placed (those with the sensor module).

    I have a question about the infrared sensor, where and how it was placed to catch from as many remote control positions as possible?
  • #4
    Juffre
    Level 12  
    Congratulations, fantastic robot. Can you know how much it costs to make such a gem?
  • #5
    sailo
    Level 9  
    dondu wrote:
    1. How did you choose the distance between the sensors and the wheel axis (center of rotation)?
    Experimentally, or specific calculations?


    We selected experimentally. This distance is especially important if straight angles are on the route. The faster they are detected, the robot will more accurately overcome this problem and will lose less speed.

    dondu wrote:
    2. What is the element connecting the sensors with the motherboard?
    Carbon fiber?


    Carbon strip.

    gbd.reg wrote:
    I would like to note that the first two drawings were inversely placed (those with the sensor module).


    Right, corrected. Thanks to :)

    gbd.reg wrote:
    I have a question about the infrared sensor, in which place and how it was placed to catch from as many positions of the pilot as possible? br >


    The sensor is located on the right side of the main PCB in front of the engine. The remote control works from virtually any angle.

    Juffre wrote:
    You can know how much it costs to make such a gem?


    The total cost is about PLN 500. I do not count the money put in prototypes and testing of individual elements. ;)
  • #6
    arti4-92
    Level 16  
    making the collars on the "main" module, which look a bit like those in hovercraft, would cause the robot to suck up to the track, of course it would result in a significant increase in grip.
  • #7
    sailo
    Level 9  
    arti4-92 wrote:
    making the flanges at the "main" module, which look a bit like those in hovercraft, would cause the robot to suck up to the track, of course it would a significant increase in adhesion.


    The flanges are made of window seals. However, there is a problem: on uneven tracks and plate connections, the robot may hang, so they can not be too long.
  • #8
    dondu
    Moderator Design Microcontrollers
    sailo wrote:
    We've chosen experimentally. This distance is especially important if straight angles are on the route. The sooner it is detected, this robot will more accurately overcome this problem and will lose less speed.

    This is only half of the restrictions, because the opposite (I think) overcoming narrow routes with too long "nose" + moment of inertia, which you paid particular attention to.

    Will you change this distance depending on the route, or did you choose the average option looking at your own experience from previous competitions?
    I ask because I see additional holes in the slats.
  • #9
    sailo
    Level 9  
    You're right, for a narrow and winding route, a shorter nose is a better solution. We have several sets of slats of different lengths and we change before the competition. The holes are leftovers after the tests.
  • #10
    dondu
    Moderator Design Microcontrollers
    And a question related to this but more about the practice of participation in such events:
    Do you do this selection before the start, seeing the route, or after passing a specific route in the test run?
    In other words, what is the practice of selecting the length of the slats to for a specific route ?
  • #11
    sailo
    Level 9  
    We choose the length before the start of the competition, although it was also possible to make a correction during testing.

    As a rule, routes in Poland are similar (gentle arches, long straights and curves at 90 degrees), which is why we usually use a variant with long slats. The exception is the T-Bot competition, which is very winding. The lines are close to each other, so the use of far-extended sensors can cause you to get out of the route (picture below).

      [Line Follower] Impact
  • #12
    konel83
    Level 15  
    Hello,
    I think so, because the discussion here shows that in the competitions you can make a trial run, did not the authors think about "remembering the route" and in the second pass use this information for optimization?

    Or even better, take a picture of the route and on a PC a program that would detect the lines and calculate the optimal speed for the engines ...
  • #13
    szymon122
    Level 38  
    I have a question: what is the name of this ribbon connecting the main PCB with the sensors. Where can you buy something like that? What are the prices? I have one with the pins removed.
  • #14
    sailo
    Level 9  
    We plan to remember the route in the next construction.

    The tape is a ffc / fpc connector, 0.5 mm pitch, available in e.g. TME
  • #15
    szymon122
    Level 38  
    I found, you can buy it for less than PLN 5? I did not know that it is so expensive; /
  • #16
    Martin_250
    Level 12  
    I warmly welcome.
    So out of pure curiosity, I would like to ask how the rules of such professions usually apply to, for example, the use of a turbine which increases the pressure and the general method of increasing pressure and what has been mentioned above, namely memorizing the route and optimizing the passage based on such data.
    This is interesting because knowing the parameters of the vehicle, ie its mass, tires or even aerodynamic characteristics, and finally the parameters of the route itself and the surface ..
    You can actually use the computer with the appropriate software to perform on-site optimization, leading to the fact that we squeeze the maximum from the vehicle.
  • #17
    arkady_pl
    Level 24  
    Martin_250 wrote:
    using the computer with the appropriate software, perform on-site optimization, leading to the fact that we will squeeze the maximum from the vehicle.

    The meaning of these professions will lose. The point of this competition is to build a robot that will negotiate a route it does not know and will do it well and quickly.
    Currently, these competitions are a software race. The structures of these best players are very similar. The winner will be the one that has better grip, greater dynamics and a better algorithm to deal with obstacles, and lately at high speeds, the speed of microcontrollers and the speed of A / D converters is becoming more and more important.

    Personally, I should say that they should introduce an additional factor to these professions: the number of transistors in the control system. For example, for each transistor the player would receive 0.01s to the overall score. Then the one who will optimize the layout will win. Because it's no art to implement mega software on fast microcontrollers.

    In a moment, there will be a robot with Full HD camera on the nose, a fast image analysis system and in flight will optimize each route. Only then will I ask: are these professions still needed? Where's the fun of building? In a moment, someone will sell blocks and software as KIT and everyone will buy one. That's probably not the point, right?

    I would introduce restrictions (exemplary) to this competition:
    - up to 500 transistors
    - up to 6 light sensors
    - maximum external dimension of the vehicle (150mm x 100mm)
    - each transistor adds 0.01s to the result
    - without clamping pressure systems (only natural gravitational pressure)

    Cool is the version of Line Follower where there are obstacles. If there were hardware limitations, then finally the players had to actually work on the tools instead of just implementing the next algorithms. Then it would be construction races and not software.

    Imagine for example that it is 1970 and you have to build a robot for the Line Follower competition. This is a challenge. Now it is very easy.
  • #18
    treker
    Level 25  
    arkady_pl wrote:
    Personally, I should say that they should introduce an additional factor to these occupations: number of transistors in the control system.

    An interesting proposition, I met her for the first time. However, I do not know if this would help in the development of competition. It must be remembered that such robotic professions are intended to encourage students and students to build something practical. Should it be limited from the very beginning?

    I even saw small children at competitions, who with their parents made a robot for Arduino. They probably would not even realize how it is with these transistors and abandoned the topic. Anyway, the electronics are moving forward, which is bad in using vision systems or fast systems? After all, everyone has the same access to them, if the key to building a better robot is to refine the algorithm, then it should be refined and that's it.
  • #19
    arkady_pl
    Level 24  
    treker wrote:
    arkady_pl wrote:
    Personally, I should say that they should introduce an additional factor for these occupations: number of transistors in the control system.

    An interesting proposition, I met her for the first time. However, I do not know if this would help in the development of competition. It must be remembered that such robotic professions are intended to encourage students and students to build something practical. Should it be limited from the outset?

    Of course, you should not limit it. I'm talking about it to make competition more ... ... exciting.

    treker wrote:
    I even saw small children at competitions, who with their parents have put together a robot for Arduino. They probably would not even realize how it is with these transistors and abandoned the topic.

    Of course, you are right. I encourage my kids by showing an easy way to build. And then I raise the bar. Because, as I wrote, building a robot from "cooked" is simple.

    treker wrote:
    Besides, the electronics are moving forward, which is wrong with the use of vision systems or high-speed circuits? After all, everyone has the same access to them, if the key to building a better robot is to refine the algorithm, then it should be refined and that's all.

    Personally, in such competitions, I see the fun of competition in the other direction, or make a simple robot that will overcome this more advanced one.

    My message is that I dream that young people optimize devices instead of complicating them more and more. He dreams and thinks in terms of what to do to do it without a computer. Of course, not always and not everything is possible or there is no point in trying to optimize because it quickly turns out that the computer is the most optimal solution, but in the case of line followers, especially those in the basic version of the track, it can be done without a computer.

    Maybe I will pass my message in a different way: until recently it was a race of constructors and electronic engineers. Currently, they have nothing to say because now Line Follower is a race of programmers. And that's what I mean the most about. This is a competition for IT professionals, not people from robotics.
  • #20
    treker
    Level 25  
    arkady_pl wrote:
    They currently have nothing to say because now Line Follower is a race of programmers. And that's what I mean the most about. This is a competition for IT specialists, not people from robotics.


    I have traveled around 10 times a year for such competitions and so far (at least in LFs) I have not met with typical computer science. They are always people interested in electronics. Everyone is trying to improve something in the robot also from the technical side.

    I am more concerned about the temporary stagnation in this competition. It used to develop faster, now there are accurate descriptions on the Internet, and some have big problems with the construction of a bearable robot.
  • #21
    arkady_pl
    Level 24  
    treker wrote:
    I am more puzzled by the temporary stagnation in this competition. It used to develop faster, now there are exact descriptions on the Internet, and some people have big problems with the construction of a lucid robot.

    Everyone should copy the SAILO user structure and then it would be people races from the PID controller configuration ;)

    You put it in a good TREKER, I write about it, that the constructions are, you can build them and many resemble or copy them. And what can be changed in such a construction? Weight, downforce, number of sensors, engine dynamics? And only the algorithm remains. The standstill will end when there is a lighter and stiffer laminate, lighter motors, lighter batteries, lighter rims and tires, lighter electronics, a stronger turbine, faster electronics.

    That is why we are stagnant. Hence my suggestions to make competition more difficult - add a new category in which the points get also for the construction. This is partly done in more advanced competition with obstacles but this only affects the software (and several sensors) and not the overall structure.

    My tots as they learned about competition with obstacles, they said: easy. Do you know what the idea of overcoming obstacles was?
    Because the obstacles are limited in size, they wanted to build a robot that would take them "astride" :D Can you imagine these "rolling" machines bypassing obstacles? :D

    Yesterday with my puppies we came up with a new Line Follower - if you could call it that. They came up with the idea to convert the helicopter so that it flew 5cm above the board and flew over the line. Oh ... and he understands, this is a fresh approach to the subject ;)
  • #22
    treker
    Level 25  
    arkady_pl wrote:
    add a new category


    There is already a division into LFy Standard and Turbo (with turbine). Amazingly, new competitions have ceased to appear in LF Turbo and the competition with turbines is dying a bit. I do not know if new competition would be accepted at the moment.

    LF games with obstacles could be more interesting if the obstacles were invented wisely. Those that are now are not overcome in a sophisticated way.
  • #23
    arkady_pl
    Level 24  
    treker wrote:
    LF games with obstacles could have been more interesting if the obstacles were invented wisely. Those that are now are not overcome in a sophisticated way.

    Oh, I strongly support your opinion.
    And if the obstacles are sophisticated, then I will not insist on millions of transistors because it will be a cybernetic task and not just the PID chasing the line with a bit of chasing the wall. Then I will be delighted when someone builds even a robot "transformer", which with the use of vision systems will overcome the obstacle course. And if we add to this, that the board will contain small water reservoirs and mud as well as bridges from two beams and false roads "ala maze" then this competition will become a structural and cybernetic challenge. Such a "biathlon" in the issue of robots. But there would be competitions!