This story is a tribute to my robot Parazit, Champion of Europe and America in 2011. Parazit has donated his parts to a new robot, But he will live forever in my heart.

One Yuko Point

Vitalij Rodnov vitalijrodnov@gmail.com www.balticrobotsumo.org www.fsi.co.jp/sumo-e

Please note, that this article was written at the beginning of 2011. Some information may seem obvious or obsolete to the readers in 2012. I left everything AS-IS, because I wanted to keep the atmosphere of my first experience in Japan. I. The Fight Night. 5 AM. I can not fall asleep. Again and again before my eyes I see red and blue rings of 22nd All Japan Robot Sumo Tournament. How many times I have endured bitterness of defeat? One minute on the Dohyo as result of two months of mad race against time. For two months all thoughts were only about a victory, every day was dedicated to a victory.

Kokugikan Hall. Human-Sumo and Robot-Sumo events take place here

And here I am on the arena, as on a razor’s edge. Parazit fights against robot named Samurai. The first fight – I was late to switch on Parazit, and he is thrown out from a Dohyo after a head-on collision. The second fight – by miracle Parazit picks up the opponent in a frontal attack. Two robots can not move a little, pushing each other and burning their tyres. Referee gives a stop signal. Parazit has lost almost all rubber from his wheels- pieces of tyres lie on the Dohyo. The third fight – Samurai starts too early and gets a warning. The fourth fight - Samurai makes mistake, Parazit counterattacks and earns one point. The score is 1:1. Last, fifth fight, kills my hope of a victory. Game over. No feelings, no pain, like after shock or trauma. I leave the arena. Huge portraits of Japanese sumo champions hang under the ceiling of Kokugikan hall. They laugh at me, I hear their voices. I shout to them: “I will be back for revenge!”

Parazit fights against robot Samurai

II. The Letter from a Dream In one usual rainy autumn day, checking my e-mail, I found the letter from David, the organizer of RoboGames. The letter said that winners of RoboGames (me, Maris Abele and Audrius Knolis) are invited to participate in 22nd All Japan Robot Sumo Tournament, the most advanced robot sumo competition on the planet. I re-read the letter again and again. I can not trust my eyes. Really, it is not a dream? I shout like a mad, frightening my wife and children. They cautiously come to my room, trying to understand, whether I have gone mad. It seems that their fears are true. I can not hold emotions inside myself. I dreamt of participation in this competition for so long time!

All Japan Robot Sumo Tournament is the Everest of robotics. Everyone dreams to conquer it, but not everyone is able to do it. Am I strong enough? What can I oppose to 20 years’ experience of Japanese robot builders? Doubts away! Long way to Fuji starts with the first step. Right away I write the reply that I accept the invitation. I wish to shout, to run, to fly. Who said what it was the usual rainy day? Probably, it is one of the best days in my life! Go forward, towards the Dream!

Interview after the match. “Next year I will come to Japan to try my luck again”

III. The Reality Next day, having cooled down from emotions, I count up the assets. Results are unfavourable. My robot Master of Disaster, the champion of Europe and America in 2010, obviously does not hold out to level of Japanese robots. Almost any Japanese sumo robot is three to ten times faster and more powerful than mine. Fastest Japanese sumo robots can reach speeds up to 5 metres per second. Sometimes fight lasts less than your eye blink. It becomes clear that I need to build a new robot, or defeat is inevitable.

...Sometimes fight lasts less than your eye blink…

I take from the shelf two brand new Maxon motors and look at them. Best Japanese robots use the motors like these. Motors are matte black, weigh half kilogram each. I was experiencing similar feeling of respect and danger when holding in my hands a weapon on Military service. How much efforts and money I have spent to get these motors! But what else do I have? Nothing. Only a name. Parazit. Yes, the name was born prior to robot itself. The start of robot building was postponed many times because of time shortage. The invitation to Japan had put business in motion. From October 20th till December 20th, every day has been devoted to creation of Parazit. My wife tolerated my hobby for a long time, but now she had seriously begun to worry. I have simply dropped out of a reality for two months. My five years old daughter has thought up to herself a new game: she turns around, waves her hands and repeats “I’m broken robot, I’m broken robot”. Obviously, one needs certain obsession in every sort of activity. I simply can not change myself. Let it be my way…

IV. The Birth Motors What did I know about Japanese sumo robots? Not too much. All my knowledge was based on youtube videos and sketchy records in Japanese blogs. First of all, it was necessary to find out, what motors use Japanese robots. Motor brand was obviously Maxon. Maxon is generally recognized as one of the best manufacturer of motors in the world. Not a surprise that many Japanese robots use Maxon motors. But what type of motors? I compared photos of robots to the documentation on motors. I had defined motor type by an arrangement of holes on motor's back sides.

Motor type on photo was identified by an arrangement of holes on the back side of the motor

The result has thrown me into confusion. Capacity of one such motor equalled 150 Watt while capacity of one motor of my robot Master of Disaster equalled only 6 Watt. Means, the average Japanese robot with two motors (2*150W=300W) was 10 times more powerful than my robot with six motors (6*6W=36W). Maxon brand is well-known not only for the high quality, but also for the high price of their products. The motor model 148866 costs $360 without taxes and shipping. Such a price is capable to frighten off many, but not such madmen as me. After long and exhausting negotiations, I have managed to get two brand new 148866 motors at a half price. Having looked at parameters of this series, it is clear that all other motors surpass model 148866 in efficiency. However, do not forget about voltage! Motor 148866 has 12 Volts nominal voltage. Other motors have much higher nominal voltage (24 to 48 Volts) that will lead to inadmissibly high weight of the batteries.

6 Watt motor (Master of Disaster) against 150 Watt motor (Parazit)

Japanese robots do not use exclusively direct current motors. One robot uses 200 Watt brushless motors with custom-made drivers. However, this power has not helped him to get to the finals. Another robot uses Mabuchi motors, but it is more like Japanese exotic. Moreover, I found the description of an upgrade, which some Japanese players perform on Maxon motors. They disassemble collector side of the motor and replace original thin brush wires with wires of bigger cross section. In theory, this upgrade raises power of the motor because more current can flow through wires with lower resistance. I did not try this upgrade so far.

Want to try this on a motor which costs 350 dollars?

Drivers How to control such a huge power? 148866 motor’s stall current is equal to 100 Amperes. It means that driver’s transistors should withstand a current of about 200 Amperes without active cooling. The double nominal current was taken for the account of overloads during a reverse of motors on full speed. I did not have an experience with such high currents. I asked Dima Snegin to help me to design driver’s printed circuit boards. We decided to use IRFS3004 transistors, each H-bridge leg having 2 transistors in parallel – total 16 transistors for two motors. For reliable switching of such powerful transistors, it has been decided to use microcircuit HIP4081A.

Driver’s printed circuit boards have turned out too bulky, because in the beginning it was planned to install radiators on every mosfet. During robot assembly, it was found out that boards do not fit into robot both because of excessive size and weight. Therefore we did not install radiators, in hope that copper layer of the boards will perform heat conducting function.

Two stacked driver boards (each for one motor) mounted on Parazit. 6N136 optocouplers on the left

I used oscilloscope and thermal imager to check and tune assembled drivers. Oscilloscope helped a lot to adjust bootstrap capacitors values and other components to achieve optimal switching speeds of mosfets. It has appeared that the main problem was poor switching characteristics of EL817 optocouplers, which were used to isolate logic and power circuitry. I urgently replaced them to high-speed optocouplers 6N136. This helped to achieve 200-300 nanosecond mosfet switching times. As it is known, short switching times considerably reduce heating of mosfets due to reduced switching power losses. It was necessary to reduce switching times to avoid an overheating and destruction of a motor driver boards. I removed anti-oscillating resistors from IRFS3004 gates, as they slowed down switching time of mosfets because of high gate capacitance value.

Yellow line- high side mosfet. Blue line- low side mosfet

Thermal imager has revealed one more problem. On a photo it is visible that after 3 minutes of work the motor overheats a lot, while mosfets remain cold. The reason of it is too low PWM frequency. Having increased PWM frequency from 2,5 kHz to 20 kHz, it was possible to reduce motor heating considerably. I have made the chart to illustrate how motor temperature depends on PWM frequency. You can see that the motor feels much better being PWMed at 20 kHz than at 2,5 kHz. This results from the fact that at low PWM frequency the motor constantly is in accelerating-braking mode and consequently heats up much more strongly. Increase of PWM frequency above 30 kHz is not recommended, as this may increase mosfet switching losses.

Batteries Initially my mistake was that I calculated capacity of the battery for the maximum mode. Each of two motors consumes 100 Amperes in stall. Hence, the battery capable constantly to give 200-250 Amperes of a current was required. The best Li-po batteries have an index 45С. It means that maximum discharge current may be calculated as rated capacity multiplied by 45. Thus we need the battery rated at least 4,4Аh (45*4,4=198 Ampere) to get 200 Amperes of continues current. I ordered two Turnigy-nano batteries (3S1P, 5Ah) from Hobbyking. However, the robot chassis has turned out more heavy than was planned. Therefore, the robot with 5Ah battery had 100 grams overweight. I urgently ordered other batteries of smaller weight (and smaller capacity). The only choice in Hobbyking was Polyquest 2,5Ah battery. It has been decided to perform the first runs with Turnigy-nano batteries, not wasting time waiting for Polyquest batteries arrival.

Tests have shown that speed of the robot was too low – about 1 meter per second. According to calculations, speed should be about 2 meters per second, but strong neodymium magnets have caused excessive decrease in speed. The only solution was to increase battery configuration from 3S to 4S. For this purpose it was necessary to divide one 3S battery into individual cells and to add one cell to another 3S battery in series. By this time Polyquest batteries have arrived. They are priced 2-3 times higher than Turnigy batteries. Advertising says that Polyquest cells are made in Korea, not in China, and therefore their quality is considerably higher. Shivering hands I have started Polyquest surgery. When the external protective cover has fallen on a floor, I saw so familiar “Made in China” marks on individual cells. A moment later I noticed one more unpleasant surprise. Individual cells were interconnected with aluminium strips by pressure method. How to solder thin aluminium strips? I quickly went to shop, bought a flux for aluminium soldering and started to work. Aluminium strips did not want to be soldered even with a special flux. Having lost hope, I put 60 Watt soldering iron onto soldering place and began to wait. As a result of very strong heating I nevertheless managed to reliably solder aluminium without having damaged the battery. The battery charger accepted self-made 4S battery and balanced all four cells. Now speed of the robot has increased to almost 2 metres per second. The further increase in speed of the robot was impossible because of weight limit. The robot weight was 2970 grams, and upgrade to 5S configuration would lead to approximately 100 grams overweight. I decided that Parazit will compete with 4S battery.

A day before competitions I bought spare Hyperion 3300mAh 4S battery in Akihabara. Parazit used Hyperion battery during 22nd All Japan Robot Sumo Tournament. Parazit uses separate power sources for logic and motors circuits. Logic is powered from small 300mAh 3S battery, and motors are powered from 3300mAh 4S battery. Logic and motor circuits are galvanically isolated by optocouplers to prevent motor noise from disturbing logic operation. Later on, after the competition, I have decided to try 6S battery configuration to get even more speed. I took two 3S 2500mAh Polyquest batteries, connected them in series, and secured them on Parazit with strong tape. With 6S configuration Parazit competed at RobotChallenge-2011 and RoboGames-2011 and won first places.

Gears I decided to start robot building from creating 3D CAD model. Photos of Japanese robots were taken as a basis. Eduardas Lukosiunas has transformed my ideas into 3D sketch. The main problem was to design self-made gearbox, because stock Maxon gearboxes were too big and could not fit into robot chassis. Gearbox design is a trivial task, but not for sumo robot, where frequent change of a direction of rotation at full speed takes place. It is very difficult (if possible at all) to calculate necessary strength margin for such gearbox. The problem becomes even more complicated if you try to make gearbox as light as possible to fit into 3kg weight limit without sacrificing endurance of gears.

First of all, calculations have been executed in MathCAD. I knew that Japanese robots can reach up to 5 meters per second speed, but I was not sure if I can control robot at such high speed. Therefore it has been decided to limited speed to 2 metres per second at 12 Volt, wheel diameter 4 centimetres, motor speed 7000 rpm and gear ratio 7:1. Gearbox has two stages. The first stage: steel gear (15T, 0.7M) on a motor shaft and brass gear (70T, 0.7M) on intermediate axis. The second stage: steel gear (13T, 1M) on the same intermediate axis and steel gear (34T, 1M) on a wheel axis. All gears were placed in the central part of the robot, and motors and wheels were placed closer to edges.

The great help in designing of a gearbox and other mechanical parts was rendered by Vadim Shtukov from Technotool. Vadim has transformed 3D robot sketch into working drawings, has helped to pick up correct materials and has given many advices on optimisation of a design of the robot. All metal parts like axes, gears, chassis and wheels have been made with professional quality on CNC machines.

Working drawings. On the right: one wheel complete gearbox assembly (top view)

According to calculations, robot speed should reach 2 meters per second, but actual speed was hardly above 1 meter per second. The only explanation of such drastic decrease in speed was presence of strong neodymium magnets, which slowed down the robot. After the competitions, looking through calculations, I have found an annoying error in the formula. Having rectified an error it became clear that actual speed of the robot almost completely coincided with calculations, and magnets only slightly influenced the speed. As a whole, the gearbox has turned out very qualitative, with low noise level. The large strength margin led to relatively high gearbox weight. I managed to reduce weight of the gears by removing some material on lathe and drilling circular holes. I have also replaced some steel parts by the aluminium ones. However a primary factor to fit robot into weight limit, was the transition from a four-wheeled to two-wheeled configuration.

Assembled chassis with 4 wheels (rear two wheels with gears were eliminated later)

Wheels Majority of Japanese sumo robots have two-wheeled configuration. This provides a great speed at turns. I have initially decided to make the robot with four wheels, with possibility of transition to two wheels, because I planned to run the robot not only on metal, but also on a non-metallic surface. For example, rings at RobotChallenge competition are made of not magnetic material, so four wheels are necessary for stability of the robot. All other competitions have magnetic ring surfaces, so two wheels with magnets provide sufficient stability to the robot. After gearbox has been made it became clear that the robot has 400 grams overweight above 3kg limit. Such a big overweight could be corrected only in one way - transition to a two-wheeled configuration.

Two-wheeled configuration

I did some test runs with four wheels, and transition to two wheels became even more apparent. After several sharp turns, rubber on all wheels has been damaged. Rubber was simply torn out of the wheels because of excessive tangential loads. Other problem was the material for the tyres. It should be very durable and provide maximum friction between wheels and ring surface. These two factors contradict each other. The harder the tyre, the more durable it is. But hard tyres slip easily, and provide worse friction compared to soft tyres. I mastered to cast self-made tyres from polyurethane of hardness ShoreA 20-30 for my robots. With such self-made tyres my robots won competitions in Europe and America in 2010. After the first tests of Parazit, it was found out that such tyres are too soft and do not stick well to aluminium hubs. Superglue (cyanoacrylate) has appeared the best solution to fix tyres on wheels. I carefully tore off tyres on both edges of a wheel, filled cavities with superglue and pressed tyres to wheels. It is necessary to try not to drop superglue on outer side of tyres as it will spoil their surface.

Shortly before competitions I found company Elastomer, which specialize on polymers. I asked them to cast two tyres for me. They have warned me that term of use of one component of the rubber has expired, and it would take too long to order fresh one. Therefore I decided to use available material. I was told that it would be almost impossible to tear off the tyre from the wheel, as they use some kind of special glue of own invention. Really, tyres from Elastomer kept on wheels very strongly. For comparison of tyres, some “burnout” type tests have been carried out. The robot was put in front of the wall and motors were switched on for 1-2 seconds at full speed. Average current was measured with ampermeter. Measurements have been put into the excel worksheet. Results say that ShoreA 50 hardness tyres perform much better than soft ones. Hardness may be increased even more, to ShoreA 70-90. One more advantage is that hard tyres can better handle overloads.

Traces from “burnout” test runs. Tyre rubber particles mixed with black Dohyo coating particles

As a gift from Elastomer, I have received a tiny bottle with special liquid, which is used by NASA Shuttles during landing to improve the coupling of wheels with a surface of a landing strip. This liquid must be put on tyres and after drying gives to tyres special characteristics. I did not manage to take full advantage of this liquid, as during head-on collision with Japanese robot my tyres have not sustained overloads and were torn to pieces. At RobotChallenge-2011 and RoboGames-2011 Parazit competed with new set of soft tyres. They were slightly damaged, but worked ok for a short time. Later on, after all competitions, I removed soft tyres and installed harder ones from Elastomer.

ShoreA 50-60 hardness tyre after “burnout” test runs

Blades Sumo robots use different tricks to get superiority over the opponent. One of the tricks is to pick up the enemy from below. Wheels of the opponent will lose coupling and it will be much easier to push it out from a ring. Usually sumo robots have an edge only in front, therefore they can be picked up easily sideways or behind. However, the fast enemy will not allow you to do this unexpectedly. Enemy will track your movements and will be ready for your attack. And then you will have a chance to test sharpness of your blade. Vismantas Masiokas has presented a blade for Parazit. It was a very sharp knife of an electric jointer with a corner of sharpening about 35-40 degrees. It seemed to me that with such knife the Parazit will be almost invincible.

Parazit’s old blade (thick one) compared to Japanese blade (thin one)

Unfortunately, I did not manage to test Parazit in fight before competition because of absence of a strong opponent. A day before competition we were allowed to test robots on a ring. Yoshimichi explained us the rules. He also brought his sumo robot. To my horror, his robot could easily pick up Parazit in front collision. It was no time to modify the blade and also no tools. By means of washers I changed an angle of slope of a blade. It has given certain effect. After this modification my blade simply rested against a blade of the opponent, but did not run onto it. However in real fight it has appeared insufficiently – from three head-on collisions Parazit could pick up the enemy only once.

As it has appeared later, Japanese players make their blades under the special order and sharpen a corner much more sharply than 40 degrees. I managed to bring home one of such blades – as a gift from Tanaka Shigeki. On irony, the Parazit fought and has lost to the robot which was designed by Tanaka Shigeki. Later on, after the Tournament, I removed old blade from Parazit and installed Japanese blade. With Japanese blade Parazit competed at RobotChallenge-2011 and RoboGames-2011 and won first places.

Japanese blade mounted on Parazit

Sensors Certainly robot needs fast sensors to be able to track opponent’s movements. For Parazit I have tried many various sensors, infra-red and ultrasonic. Ultrasonic sensors Maxbotix and Devantech have a wide field of view, but their response time is too long – 50 milliseconds and more. In 50 milliseconds robot passes 10 centimetres at 2 meters per second speed. Widely used Sharp GP2D12 infra-red sensors have a narrow field of view and response time around 40 milliseconds. The fastest sensor Sharp GP2Y0D340K has response time around 7 milliseconds. This is much better, but nevertheless it is not enough. Besides, visibility range of 340K sensors is only 40 centimetres. Such range is good for minisumo, but not for 3 kg sumo robots.

For Parazit it was necessary to find much faster sensors. Industrial sensors have response time about 1-2 milliseconds. It means, while Maxbotix does one measurement, the industrial sensor will make 50 measurements. The only drawback of such sensors is their high price. However miracles happen, and sometimes you can buy good industrial sensors on eBay for the very good price. I bought 10 centimetres range diffusive Sick sensors from the Israeli seller for 17 dollars each, and 100 centimetres range diffusive Keyence sensors from Turkish seller for 25 dollars each.

Red light of the side sensor. Upper you can see yellow sensitivity adjustment potentiometer of the line sensor

In total, Parazit has 8 sensors - 6 for detection of the opponent (100 cm range) and 2 for line detection (10 cm range). All sensors are PNP type. I would prefer NPN open collector, but on eBay you do not have too much choice. I have put line sensors under the front inclined plane so that they could see a line as soon as possible. The distance from a sensor to a ring is around 5см, and sensitivity adjustment on the sensor allowed me to adjust the operation moment precisely. Such arrangement allows me to use sensors of the usual size. Industrial sensors are much less subject to influence of light disturbances. Therefore they could precisely detect a line and they do not react to scratches on a ring or bright external illumination. Three opponent detection sensors have been directed forward, one sensor 90 degrees left, one sensor 90 degrees right and one – back. Visibility range of these sensors is 100 centimetres. It is quite enough for a ring diameter of 154 centimetres.

Line sensor (looking down) and side sensor mounted on Parazit

All sensors are powered from the 300mAh microcontroller battery. It means that the output signal of a sensor changes from 0 to 12.5 Volt. It was necessary to use simple voltage dividers to connect these signals to microcontroller inputs. NPN sensors with an open collector output can be connected to the microcontroller directly without any dividers. You need a pull-up resistor from sensor output to 5V supply of microcontroller. However if you find a suitable sensor on eBay, it is not important what type it is – PNP or NPN, you just grab it.

Three front sensors on Parazit. On the upper left there is a hole for ISP programming plug

There are two types of diffusive sensors- with infrared light beam and with visible red light beam. One useful property of infrared light beam sensors - many of them emit an additional beam of visible red light for targeting and aligning purposes. This beam is well visible and facilitates adjustment of a direction of a sensor. On Parazit these red beams also serve as indication of that a microcontroller power is switched on.

NPN open collector and PNP open collector sensors connection to microcontroller, for example ATmega

Magnets Magnets are one of the most crucial parts of Japanese robots. Magnets presses chassis against Dohyo surface, which ensures better friction and high pushing power of robots. Magnetic force can reach 100 and even 200 kilograms. It means robot weigh on the Dohyo is not 3 kilograms, but may be 103 or even 203 kilograms. Can you imagine how much power do you need to push such robot out of the ring? Magnets also ensure that blade is firmly pressed against Dohyo.

Big magnets saturate thin steel easily. It’s better to use medium to small size magnets

Many robots use open magnets, some use closed systems. Closed magnetic field is more efficient, because magnet size can be reduced without decreasing of magnetic force. Firstly Parazit used big fat “cup” type closed magnets, of diameter 42 millimetres, rated force 50 kilograms each. These magnets worked good on our training ring, 3 millimetres thick. But on Japanese ring, which is only 1.6 millimetres thick, these fat magnets have saturated steel ring surface and magnetic force was decreased a lot.

Big fat magnets mounted on Parazit

You must place magnets as close as possible to the ring surface, because magnetic force decreases very rapidly with distance. It is a good idea to cover magnets with Teflon film, and let them slide on the ring. This will minimize distance between magnets and ring, and also minimize friction losses. Try to replace worn out Teflon pieces as soon as possible, because metal-metal friction is much higher than Teflon-metal friction. Worn out places will make your robot slower.

Small magnets coated with Teflon tape to minimize friction losses

You can easily attach “countersunk cup” type magnets to the robot with bolts and nuts. Other types of magnets are not so easy to attach. Hot glue or superglue probably will not work. Later on, after the Tournament, I removed fat magnets and installed smaller magnets, diameter 20 millimetres, rated force 10 kilograms each. Smaller magnets do not saturate thin steel and also allow to be placed around wheels to distribute magnetic force more evenly around bottom of the robot.

Open magnets on Japanese radio-controlled robot

Logic In the beginning I wanted to use ATmega8 microcontroller in SMD package to reduce the sizes of a board. But later I decided to use ATmega8 DIP package as it allows fast replacement of the controller in case of suspicion on malfunction. Unfortunately, this tactics has not proved to be true. Fights pass in such prompt rate that you have no time for repairs. So any breakage or a program glitch practically means loss. It was decided not to bother with strategies, preferring a brutal frontal attack. Oh, how I was wrong! Weak European opponents have created an illusion of brute-force efficiency for me. But Japanese robots were much smarter. The rare Japanese robot attacks frontally. Most fast maneuvering, trying to go to the enemy side or rear. Conversely, some robots wait until the enemy comes closer, and attack at precisely selected moment, like a dagger punch, throwing opponent out of the ring.

Typical Japanese robot with side blades

Many robots use flags to confuse the enemy. As one Japanese player told me, almost all robots measure width of the enemy and aim at the centre, not to be fooled by flags. The winner of competitions used flags even more creatively. In a final fight he has jammed the left flag, so only the right flag has fallen down. And the opponent, incorrectly having calculated the robot centre, has attacked on a wrong trajectory, has been picked up and thrown out from a ring like a kitten. Having seen this, the Kokugikan hall has simply blown up with shouts and applause!

Final fight for the first place: TeamQ robot Sakura was fooled by flags of robot Six Dimensional K

Later on, after the Tournament, I removed custom microcontroller board and installed Pololu Baby Orangutan board with ATmega328P. This did not give me any particular advantage, because entire program was less than 8 Kbytes. But it simplified wiring and was smaller than my custom board. I also installed rotary switch for strategy selection. First, when referee gives a signal, robots must be placed on the Dohyo simultaneously. Then you have time to select strategy and make other preparations. Then both players report to referee “Ready”. Fight is started when referee gives another signal. Usually referee says [HAKKIA NOKOTTA], which means “Stay on the Dohyo”, or just [ICHI NI SAN], which means “One, Two, Three”

Pololu Baby Orangutan board installed on Parazit. Rotary switch is on the right, mounted vertically with ISP plug

V. The Human factor Hiroshi Nozawa, the president of company FujiSoft, is the founder and constant sponsor of All Japan Robot Sumo Tournament. Event takes place once in a year, and we participated in 22nd competition. During his speech at opening ceremony, Hiroshi Nozawa stressed that All Japan Robot Sumo Tournament must become International event and international players are welcome. Of course, players must deserve personal invitations by winning well recognized robot sumo events around the world.

There are two categories of robots - Autonomous and Radio-Controlled. Although the RC gradually transforms into semi-autonomous, as the microcontroller on board the robot does not allow the operator to make critical errors. For example, if the operator will direct the robot on the white line, the microcontroller takes over control, turns robot away from the line and returns control to the operator, even though operator simply press “forward” button on remote control unit. But despite these tricks, I must admit that Japanese RC operators are extremely fast and dexterous, on speed of reaction probably coming nearer to lifeless electronics of Autonomous robots.

Two radio-controlled robots ready for the fight

Japanese people are well known for their love to technology. As I was told, practically at every school there is a local club, where students build a variety of devices (not necessarily related to robotics). Local robot sumo competitions are called Pre-Tournaments, and winners of Pre-Tournaments get invitations to participate in All Japan Robot Sumo Tournament, so-called “Grand Finals”. Each participant, even eliminated in first circle of Grand Finals, gets the award of 30000 yens. Further awards increase, and the absolute winner receives about 12000 dollars, the certificate of honour, a big shiny cup and a photo with organizers and guests of honour.

BalticRobotSumo is recognized as one of the most important robot sumo events in the world

Ah, those beautiful cups, made in the form of man, letting out a bird from his hands in the sky. The great symbolism is hidden in the cups design - seek forward, upward, like a bird, not afraid of height and not thinking about the falling! Let out your dream! Let it fly in the sky! Conquer the highest peaks! Fly on wings of your dream! During the opening ceremony all participants were seated on stage. The beautiful cups were just five metres away from me. The stress was so high that I began persuade cups, any of three, to become mine. At home, when I told my friends about my conversation with cups, one said: "Well, it is clear why none of the cups become yours. You were talking to Japanese cups in Russian, but they just do not understand Russian!"

Head referee Mr. Kitano (on the left) holds beautiful trophy for the winner

Sitting in a bar after competition, I asked Japanese sumo referee: “What is the strength of Japanese robots? Powerful motors? Sharp blade? High speed? Smart strategies? All-seeing sensors?” The answer was much shorter than the question: “Balance between all you’ve mentioned!” It made an impression. So, that gave the name of my new sumo robot. I called it Melody. Laugh, but I like it. You can find balance, or harmony, in the Mother-nature, or in the music. Yes, music can do wonders. I hope, my Melody will sound great at next All Japan Robot Sumo Tournament.

CAD model of my new robot Melody

VI. NOT an epilogue Definitely it is NOT, because epilogue means termination, the stop, the end. For us it is only beginning, the new step, new level of sumo robot building in Europe. We were the first Europeans who have brought home real experience of fights with Japanese sumo robots, though with a bitterness of defeat.

It’s always the beginning… Even after loss… Stand up and fight again!

Gained unmatched experience has allowed Parazit to take the first places at largest European (www.robotchallenge.org) and USA (www.robogames.net) competitions in 2011. Our motto: Only strong opponents can make us stronger. Melody is already invited to 23rd All Japan Robot Sumo Tournament, which will take place in December 2011. The winners of BalticRobotSumo (www.balticrobotsumo.org) also got personal invitations to Japan. This is a great honour for us and a sign of recognition of our successful activities of promoting robot sumo outside Japan. To all readers who have read this story to the end, I wish one:

LET YOUR DREAM TO FLY!