IR ball modulation change proposal

RoboCupJunior Soccer
1

Hello, I think that we should change the IR ball modulation standard from mode A to mode D. For multiple reasons.

  1. The original intention for using dimed pulses was to have some way of detecting the approximate distance from a ball using standard sensors such as TSOP1138 in the way that a higher distance meant fewer pulses would be detected (8-12-16-20). But in reality, because of insufficient uniformity of IR radiation from the ball and the non-linear part of the sensor pulse weight corresponding to irradiation distance information gained this way becomes unuseful and teams are only using direction information. When we change official modulation to mode D getting direction would work the same way as before and no modification would be needed.

  2. There is also a problem with IR light bouncing from obstacles and walls so in the end there are multiple ‘paths’ from the IR ball to the sensor which is even more confusing for the sensor. When using mode D as the official way of modulation it would mean that all ‘dimed pulses’ are reflections and should be ignored when now reflection and dimed pulse could be easily mistaken when using mode A.

  3. Using mode D would create a simpler path for understanding how this detection works for new teams and would also make a more easy path for new teams. Also would make new designs of the ball simpler and less expensive.

  4. It would create much more precise detection of distance/angle when using photodiodes with ADCs because it would make the detection algorithm more accurate due to better decisions on what is reflected only on comparing it with max intensity.

  5. It would not require any hardware changes to existing robots(in basically all cases) and would not require the purchase of new hardware/balls (just switch a switch on a ball).

  6. It would work with all most commonly used sensors such as IR seeker V2 and all sensors based on TSOPsXXX

Of course, this change would not take place for this season.

image
image2383×1177 415 KB

image
image738×841 38.9 KB

2

Hello,

Thank you for sharing your idea. Many teams can accurately determine the distance from the ball. Additionally, the challenge of estimating the ball’s distance helps students become familiar with various modulations. I use this topic for educational purposes.

All the issues you mentioned are related to the sensors you are using. First, you need to understand the problem. Most sensors used for ball detection have automatic pulse adaptation. This means that as soon as they detect a signal, they adjust themselves based on the received intensity and change their gain, called AGC(Automatic Gain Control)

Automatic gain control (AGC) is a closed-loop feedback regulating circuit in an amplifier or chain of amplifiers, the purpose of which is to maintain a suitable signal amplitude at its output, despite variation of the signal amplitude at the input. The average or peak output signal level is used to dynamically adjust the gain of the amplifiers, enabling the circuit to work satisfactorily with a greater range of input signal levels.

These sensors are typically used in remote controls. Consequently, this function allows them to detect a weak pulse from a remote control at a distance of approximately 45 meters. Therefore, you receive reflections from the ball, but estimating the distance from the ball may be challenging for you.

To solve this issue, there are several solutions:

  1. Use sensors without AGC: Some older sensors that lack AGC are still available in the market, although finding them can be difficult.
  2. Apply a 38KHz filter to a regular IR sensor: This method is usually not cost-effective.
  3. Limit the field of view significantly for sensors like TSOP1138: This allows the sensor to use maximum gain, enabling you to estimate the distance. To achieve this, fully cover the sensor’s field of view and then open only a very small hole in front of it. Keep in mind that these sensors are incredibly powerful. You might think this is impossible, but I have tested sensors that can detect even the tiniest openings. Therefore, first, ensure that your sensor’s field of view is properly restricted.

I hope that using these methods will help resolve your issue.
Feel free to ask any questions or clarify any doubts.

Best wishes! :blush:

3

Hello,

first of all, thank you for your response, I really appreciate it.

“Many teams can accurately determine the distance from the ball.”

First I wanted to ask about how much accuracy we are talking about in terms of distance and what exact type of sensor/method/technique is used. When it comes to determining the angle I am sure that many teams can easily determine great accuracy but in terms of distance, I am not that sure not only because there is great fluctuation of IR light intensity while rotating a ball/ ball being placed in different angles as well as in the terms of IR reflection on the field.

"All the issues you mentioned are related to the sensors you are using. Use sensors without AGC : Some older sensors that lack AGC are still available in the market, although finding them can be difficult. "

First of all, to recommend teams use sensors that are hard to get, and in most cases are not produced anymore is very bad for the future of this competition and is not in line with the idea of a lightweight league, where detecting a ball should be really easy and cheap and accessible (not need for the camera). Also when we recommend using such sensors we will “kill” new teams even before they start doing lightweight because they will not be able to get an idea of how to detect the ball quickly and easily.

Another thing about using sensors without AGC is when using such a sensor, IR light intensity is directly responsible for the change in the output so where do you see the need for dimed pulses alongside full-intensity ones?

"Apply a 38KHz filter to a regular IR sensor : This method is usually not cost-effective."

This is not true because these days many MCUs can read this signal directly with ADCs using photodiodes/phototransistors for low prices such as STM32G474CBT6 STMicroelectronics | C730123 - LCSC Electronics and more… Also there are other analogue ways of doing this same thing but there are in the present outdated and not as accurate or cost/space effective as using this approach.

"Limit the field of view significantly for sensors like TSOP1138 : This allows the sensor to use maximum gain, enabling you to estimate the distance. To achieve this, fully cover the sensor’s field of view and then open only a very small hole in front of it. Keep in mind that these sensors are incredibly powerful. You might think this is impossible, but I have tested sensors that can detect even the tiniest openings. Therefore, first, ensure that your sensor’s field of view is properly restricted."

This is the most common approach that most of the teams use, what this does is basically use the first portion of the graf I posted in the previous post and therefore ensures a more linear change of the output. But still using this approach is as well not dependent on dimed pulses and would work with mode D as well.

But big tumps up for being a mentor and using this topic for educational purposes. If you have some educational materials I think that there are multiple people that will benefit from it (even when it might be in a different language) so if you could share it we would be grateful.

Thanks again for your opinion and if you need clarification on any point made please do not hesitate to ask

Best wishes!

4

Also one of the original intentions was to be able to detect ball easily with common sensors. Here are some documents from the original introduction of this type of modulation.
https://drive.google.com/drive/folders/18GV9xv1I0IeJWxWjoTU4oSMSzOTz13s8?usp=drive_link

5

Hi,

I like the idea, especially of making it easier for newer teams.
Maybe we could try this modulation as a technical challenge at the Dutch World Open and ask if teams find it better to work with.

1 Like
6

Hi,
I agree with Elias,
what about testing it in Eindhoven? We could also test it in Hannover at the European Open.
We have to consider one more thing:
The original ball is discontinued and hard to get. We have to think about something new.
(my suggestion would be something like a golfball sized IR ball with only two 180 degree IR diodes but with a “milky” body, so that it “glows” as one thing and does not send multiple spotlights)
However, designing this new ball with only “mode D” could be easier and cheaper as well.

7

Hi,
I already did developed and ordered new balls for testing in Eindhoven and hopefully will be able to get them to European Open as well.

image
image886×891 191 KB

Also when it comes to golf ball sized IR ball I have prototype on the way so hopefully we will be able to test them, although as big change as switching ball size take some time so for this reason I did developed normal sized ball.

That is general idea but 2 LEDs are not enough because even when you get 180 that does not mean that it it homogeneous on 180 degrees when it comes to light intensity, it looks something more like this

image
image667×642 42 KB

so what you relay want is to have as many wide angle as possible with god light dispersion on shell so it is as homogeneous as possible.

Also why I am talking about also changing modulation is that if we want a lot of them cheaply using something like “mode D” would simplify schematic from something like this https://tinyurl.com/ys8j6dxv to something like this https://tinyurl.com/2xg8b49o. So it would be smaller(be able to fit in golf ball easily,also would save some battery that is in the case of golf ball crucial).

8

Hi Jakub,
great work. :clap: :clap: :clap:

Is this the ball we have seen in Varazdin last year?
Probably a “milky” ball shell could help making the light intensity more homogeneous?

Hope to see you in Hannover.
Roland

9

Hi Roland, it is new optimized design.

Probably a “milky” ball shell could help making the light intensity more homogeneous?

Yes it will be “milky” but will try few different materials for injection molding (also still some 3d printing variants I have in mind for testing) also some “lenses” to make it as homogeneous as possible.

Hope to see you as well :+1:
Jakub

10

Just to clarify. EK Japan’s IR balls are still in production. They are hard to get since you need to directly contact the company to purchase now. But if there is any bulk order, I can help initiate the purchase.

11

Hi Jakub, thanks for putting so much effort into a new ball.

As Amy said, there is a way to still get the “old” ball.

So it would make sense to concentrate on a new golfball-sized IR ball.
We could think about changing the ball size for 2vs2 lightweight and leave the old ball for “1vs1”.
At least for the first couple of years.

As this would be a mayor change we are talking about a rule change suggestion for 2027.
This means, we should have a new ball in Summer 2026 available to buy.
Do you plan this ball to be opensource?
Do you have partners for building 1000+ balls?

Greetings

12

The old balls are also very expensive and break relatively easily, we bought new ones last year, which are already broken (broken cables & capacitors) and they only get used once a year (so two times by now).

So a cheaper and easier to get alternative would be much better than relying on the old ball & manufacturers.

Also, a golf ball sized IR-Ball is a little bit so small in my opinion, we should change the LWL rules in terms of size, but it should still be larger than the OL, because of building cheap and proficient robots for 18cm is hard and expensive.

13

Yes it will be fully opensourced.

Yes, a lot of tough has gone into making it not only easy to manufacture them but also easy to get.
The details are still being finalized but I’ll certainly update the forum once we are done with preparation.

I must say that it is actually really good design by EK JAPAN but it is little outdated to this point and this brings some issues.

I think that using golf ball size for IR will bring many advantages such as teams will be able to develop and build upon designs of dribblers from open/ small size to major league. Also, goals sizes can be made to fit just one type as well as better homogeneity of light, lower price, easier shipping also more complex strategies can be made, smaller kicker space needed less powerful kicker (more lightweight).

When it comes to 18cm, this change was made so there was more space on the field for actual gameplay so there is always this fine balance between field size and robot size: when you try to minimize robot size so it is still easy to build and maximizing field size so it is easy to transport/manage. I think that 22 cm for light weight is optimal and shouldn’t be changed . 18cm is still possible but requires more engineering and resources but it helps when encouraging teams to make more sophisticated strategies as ball passing because you have more space for movement and robots are closer to players than obstacles.

14

If I understand this correctly, switching to Mode D would mean that we will not be able to digitally count the number of signal ups and downs to determine the rough distance. Instead, we would have to use ADC to measure the IR signal strength on IR photo sensors, and then compare them to find the max in order to determine the distance. Is that how you expect it to work?

Without actually trying it, I am not sure if this will have any negative effect on the accuracy of direction determination. But I am certain that this will cause significant changes to the robot receiver’s circuitry and firmware. First of all, the kind of IR sensor/Photo sensor used will be different. Secondly, the MCU’s GPIO ports used may need to be different depending on the MCU used. Thirdly, firmware will need to updated. I personally am not fond of this proposed change.

15

Hi,
all reasons for this change are here:

and for your concerns

Without actually trying it, I am not sure if this will have any negative effect on the accuracy of direction determination. But I am certain that this will cause significant changes to the robot receiver’s circuitry and firmware. First of all, the kind of IR sensor/Photo sensor used will be different.

To be able to detect ball direction you can still use TSOPs approach as well as old IR LEGO seekers with higher precision than old ball (more LEDs, smaller size, still 40kHz based)

Secondly, the MCU’s GPIO ports used may need to be different depending on the MCU used. Thirdly, firmware will need to updated.

Updating your robots firmware/hardware is esencial for learning and it helps with teams that bring same robot every year.

16

Full disclosure. We are a vendor that developed this new IR Seeker sensor: https://igg.me/at/irseeker/ We were able to make our IR Seeker accurate by using the digital approach. We use a number of TSOP34840s to capture the signal strength from each direction. For each TSOP, we measure the width of the inverted pulse to determine signal strength. Our MCU takes many samples and uses running averages of these measurements to calculate the final strength values to return to the robot. To find the distance, we basically compare the signal strengths from all these IR receivers, and find the maximum one.

Obviously, this approach heavily relies on the steps in the waveform to determine signal strengths. If the waveform of the IR Ball changes to a non-stepped form, I guess we pretty much have to change both our hardware and firmware.

18

Hi Joseph,

I would like to make it clear that the new ball proposal would only apply to the Lightweight and Open Leagues of RoboCupJunior. We currently, nor for the foreseeable future intend to change the ball used for Entry Leagues, (typically that of Lego EV3 or SPIKE systems).

The EK Japan ball is a great size and still widely used ball, excellent for beginner teams. With it’s larger size, it is typically much easier to design a robot for as a beginner (and hence our stance).

Regards,
William

19

Hi Will,

That’s great to know. Thank you.

Joseph