Suggestions for making an inspection robot about 2 cm tall with "compressible" wheels

Posted by: @adrobot

↑

Hi, I need to make a small rc controlled device to inspect the cement screed under the roof tiles, taking into account that at the visible spot in the photos, the maximum height may be a little less than 2 cm. See Photo 1 and Photo 2 for details about the places to inspect.

I have selected these components:

1) Esp-32 Cam + OV2640 Cam

2) DRV8833 Dual Motor Driver

3) 2 x GA12-N20 6V 50 rpm motor

4) 2x 20mm Sponge Wheel

5) A flat LiPo battery …

6) I don’t know if for the back it is better to use a ball caster, a tail wheel or some other component ...

 

I’ve chosen high-torque, low rpm motors paired with soft sponge wheels, which I hope should be powerful enough to compress the sponge should the wheels touch the tile at the critical point about 2 cm high.

Can you give me your opinion about the feasibility of this solution and the choice of components ?

Thank You.

Those components appear to be viable. My concern is that the 2cm clearance is constant, in other words is there any one or more places that it is substantially less than 2cm so that the robot gets stuck. I have seen a design made for this environment that looks like a snake. Have you tested if WiFi will penetrate the tiles?

@adrobot

I live in Canada and tile rooves aren't very popular here, so the closest I've ever come to one is viewing it from the sidewalk. So, judge my suggestion accordingly..

I assume that the function will be to introduce the wee beastie from the side of the roof, always underneath the tiles, and have it make its way across a row of tiles judging the clearance of each tile before it passes below it. If there's a gap of less than 20 mm, then your robot will attempt to roll on and have the flexible rims of the wheels compress to allow the 'bot to continue. 

While operating, I assume that you'll be monitoring both your path and the tile's altitude gap continuously and both visual and moron control data will be flowing via some radio signals. You have not stated if you will require some kind of marker to be recorded noting an unacceptable reading at distance (eg 15mm gap at 3.65 metres), but I would expect such a annotation would be required if you wanted to minimize the disturbance of roofing material repairing the deficiencies.

You might consider using a borescope (camera snake) type video tool instead of an onboard camera, and a trailing wire harness to drive the motor and wheels.

The advantages are that you won't have to rely on the RC signal in noisy environments and, if your robot gets stuck you're able to pull on the camera and wire harness to apply extra force to free the robot if jammed somewhere. Also, you can mark the cable (or attach a measured cord) to facilitate determining the distance from the edge where the jamming happened or unacceptable gap is found.

I believe that determining location would be challenging on an RC robot.

It's an interesting problem. It would be tempting to make the carriage lower (as 1.5 cm wheels) on 6 wheels and break the carriage into two parts, hinged at the middle so it would better be able to navigate over the hills and valleys in the screed. However, I believe that your method of using only two wheels is superior in that it doesn't have to accommodate very rough terrain.

One last thought is that the borescope camera should not be permanently attached to the carriage. If the motor battery and controls are kept with the borescope viewer, then if the robot gets stuck, a strong pull on the borescope cord and wiring harness will retrieve them. If the borescope camera isn't very firmly attached, it'll slip off and come free. The only parts lost (if the owner refuses to let you scramble over the roof to lift the tiles and fetch the robot) will be the frame, wheels and motors. There will be NO DANGER of a sun-heated LiPo battery pack possibly starting a fire on/in the roof.

I'm curious how you intend to estimate the gap from a camera ? Will you engrave a scale, have a feeler gauge, guess, compare it to a fixed length reference rod mounted on the chassis ?

Sounds like a cool project!!!  I'll be following if you keep posting progress.  I was also wondering... about just using a bore scope.  It seems like plumbers have very long ones and they can point the head somehow and I gather can direct it down various branches.

I'm wondering... do you expect to have any 12:12 or steeper?  I think a wheeled device will be problematic if so.

Posted by: @adrobot

↑

6) I don’t know if for the back it is better to use a ball caster, a tail wheel or some other component ...

I had a little zero turn robot that I 3D printed.  I just used a replaceable skid of plastic.  It worked fine over hard floors, carpet, concrete sidewalks.  I wouldn't think it'd wear out all that quickly.  At 2 or 3 cents worth of plastic and with no moving parts to jam, get caught... it is the ultimate KISS principle.  If you use Nylon or Polycarbonate, they have excellent ware characteristics.  Doubt you have to change it out with miles of travel over side-walk rough concrete.

(I know nothing about these kind of rooves) I can't tell from the photos how rough the terrain is under them.  I'd think if roofers are slopping some kind of mortar around, they wouldn't be all that neat about it and any little blob, could high-center your robot.  Based on 2cm wheels and the thickness of your motor, you're looking at anything over 5mm could high center it.  But maybe, I'm just not visualizing the terrain that well.

Hello everyone and thank you for your answers !

My answers below:

Posted by: @zander

↑

Those components appear to be viable. My concern is that the 2cm clearance is constant, in other words is there any one or more places that it is substantially less than 2cm so that the robot gets stuck.

I cannot rule out that there are also points where the distance is less than 2 cm. I have taken measurements at several different points and the distance is always around 2 cm.

Posted by: @zander

↑

I have seen a design made for this environment that looks like a snake.

I had already seen the storm drain inspection vehicles and pipe inspection robots, but there are no such small models for my needs.

I have done numerous searches and have not been able to find anything available ready-made that is similar to what I want to build.

There are similar devices, but the main limitation is that they are too tall to fit in that small space under the tiles.

Posted by: @zander

↑

Have you tested if WiFi will penetrate the tiles?

No, but I don't think it could be a problem, because I could put myself directly under the device if there were problems with signal reception.

Posted by: @will

↑

@adrobot You have not stated if you will require some kind of marker to be recorded noting an unacceptable reading at distance (eg 15mm gap at 3.65 metres)

No special marker is required, I can simply record a video with the onboard camera and count the tiles I was able to pass to get an accurate indication of where the device is and record any critical points.

Posted by: @will

↑

You might consider using a borescope (camera snake) type video tool instead of an onboard camera, and a trailing wire harness to drive the motor and wheels.

I already have a rigid probe 15 meters long and 5.5 mm in diameter, albeit not motorized, but in this case it was possible to push it, with much difficulty, to a maximum of about 4.5 meters.

Posted by: @will

↑

It's an interesting problem. It would be tempting to make the carriage lower (as 1.5 cm wheels) ...

I've been thinking the same thing: replace the 10 mm motors with 6 mm diameter ones and 16 mm diameter wheels. This would give a good margin from the critical point, while avoiding using force to compress the sponge wheels ...

Posted by: @will

↑

One last thought is that the borescope camera should not be permanently attached to the carriage. If the motor battery and controls are kept with the borescope viewer, then if the robot gets stuck, a strong pull on the borescope cord and wiring harness will retrieve them. If the borescope camera isn't very firmly attached, it'll slip off and come free. The only parts lost (if the owner refuses to let you scramble over the roof to lift the tiles and fetch the robot) will be the frame, wheels and motors. There will be NO DANGER of a sun-heated LiPo battery pack possibly starting a fire on/in the roof.

I too had thought of something similar, but the main problem is the weight of the cable, which would have to be dragged for about 15 meters and could easily get stuck along the way. It would need very powerful and big motors ...

Posted by: @inq

↑

Sounds like a cool project!!!  I'll be following if you keep posting progress.  I was also wondering... about just using a bore scope.  It seems like plumbers have very long ones and they can point the head somehow and I gather can direct it down various branches.

I mentioned something above about the borescope: I have several probes, but the main problem is that after a few meters the weight of the cable and the irregularities of the channel prevent pushing the probe further. The most I have been able to push a probe is about 4.5 meters, but it is really very difficult because the probe tends to get stuck on the sides. In this case I would have to go in about 13 meters.

Posted by: @inq

↑

I'm wondering... do you expect to have any 12:12 or steeper?  I think a wheeled device will be problematic if so.

The pitch of the roof is about 30 degrees, but the shape and positioning of the tiles allow the radio-controlled device to move only following the direction of the ridge or eave line, and not from the eave line to the ridge line or vice versa.

Posted by: @inq

↑

I can't tell from the photos how rough the terrain is under them.  I'd think if roofers are slopping some kind of mortar around, they wouldn't be all that neat about it and any little blob, could high-center your robot.  Based on 2cm wheels and the thickness of your motor, you're looking at anything over 5mm could high center it.  But maybe, I'm just not visualizing the terrain that well.

You are absolutely right: there is a real risk that any debris such as lumps of mortar or anything else could be an insurmountable obstacle or even the device could get stuck and lost.

Anyway, the components to make the whole thing are very cheap, and with 15, 20 dollars you can buy everything, so I would like to give it a try ...

Posted by: @adrobot

↑

The pitch of the roof is about 30 degrees, but the shape and positioning of the tiles allow the radio-controlled device to move only following the direction of the ridge or eave line, and not from the eave line to the ridge line or vice versa.

Oh... I somehow got the impression you might be making this into a product for contractors or as a service and would need to handle all kinds of roof pitches.  

Another question...

What kind of maximum speed are you thinking about?  

Edit:  You gave me enough info in your part selection = ~10 feet/minute.

Hi @adrobot,

 A semi-random technical suggestion regarding the possibility of using a wire to power your robot, and maybe even used to give it a yank, if it gets stuck.

As already pointed out, this would mean that there would be no chance of battery being trapped in the roof as a fire hazard, as well as reducing the robot payload. I am suggesting the need for a buck regulator on the robot, but I am guessing that will be much smaller than the battery it is effectively replacing.

You may need to do some seaching, but you might be able to find some lightweight cable that is light and flexible enough to be pulled by your robot, without having to make the motors bigger, etc. , albeit it might still get snagged on something.

To minimise the wire requirements, I'll assume it will only be used to transmit power, and that all of the control and video are carried on WiFi or similar, as you have indicated.

Of course, lightweight cables imply thin conductors, which will reduce the current they can carry.

However, a trick to reduce this problem is to transmit the power at a higher voltage ... that is why the long distance electrical distribution is done at voltages like 400kV ... but the same principle can be applied at more modest voltages as well.

I'll give a very simple worked example with round number voltages and currents, so that you can repeat the calculation for your own case.

Assume the robot needs 5V at 1A to power it. But your wire is only suitable to carry about 0.2A. You have a power supply that can supply 5V at up to 1A.

(You might also get a 25 V power supply of course, but I have put a boost converter into the scheme to demonstrate the point. Also, if it is difficult to get mains power to the location, you might decide to use a battery to power the system, but keep the battery at the 'launch point' where you can reach it, rather than fixed to the robot.)

Note 5V at 1A, means a power of 5 (V) * 1 (A) = 5W

Use a boost converter to convert power from supply from 5V to 25V.

You need 5W of power, but at 25 V, the current for 5 W = 5 (W) / 25 (V) = 0.2 A

   which is within the capacity of your thin wire.

Then at the robot, use a buck converter, to bring the voltage back to 5V ... and given "100% efficiency" converters and "no wire loss",  the 25 V at 0.2 A would be converted back to 5 V and 1 A.

-------

Of course, 100% converters and wire with no loss are both unobtainable .. converters are typically 85-95% efficient, and all wire has resistance, so you will need to put more power into the system, and uprate everything to allow for the losses.

However, there is a second advantage to this approach. Imagine you tried to transmit 1A at 5V directly. 15 metres of thin cable would have an appreciable resistance, and the voltage at the robot end would be less than 5V, the amount less depending on both the resistance of the wire and the current draw at any one moment, given that it will fluctuate with motor speed, etc. You could try increasing the voltage from the power supply to allow for the losses, but when the current demand is low, the voltage at the robot end may go high enough to damage something.

By having a buck regulator at the robot end, assuming it is chosen appropriately, it can accommodate the voltage fluctuations on its input, and provide a steady voltage of (say) 5V at the output.

-------------

Although this message is already quite long, it only touches on many of the details. And as with all 'brainstorm' suggestions, it may prove impractical when all of the project details are considered.

If the scheme looks useful, then please feel free to ask further questions.

Best wishes and good luck with your project, Dave

I apologize for the late reply, but notifications of new posts were forwarded to the wrong mail folder ...

Posted by: @inq

↑

What kind of maximum speed are you thinking about?  

Edit:  You gave me enough info in your part selection = ~10 feet/minute.

Considering the limitation of the place where the device will move, it is necessary to prefer a low speed of movement and high torque motors.

I may add that the device should travel about 15 meters on the outward journey and as much on the return.

Posted by: @davee

↑

Hi @adrobot, A semi-random technical suggestion regarding the possibility of using a wire to power your robot, and maybe even used to give it a yank, if it gets stuck.

Hi @DavE. Thank you very much for the valuable advice, it will come in handy for future developments.

At present, I would like to develop a battery-powered device, for one main reason: there is a very high probability that the cable, even if it is light and thin, may get stuck along the way.

Posted by: @davee

↑

To minimise the wire requirements, I'll assume it will only be used to transmit power, and that all of the control and video are carried on WiFi or similar, as you have indicated.

I agree, it would be the easiest modification to transform the device from battery power to cable power.

Another development might consist of modifying a cheap endoscope for sale on Aliexpress, adding wheels to it that are driven and powered by the same usb cable where the video signal travels ...

Posted by: @davee

↑

Best wishes and good luck with your project, Dave

Thank you!

Having done further evaluation on this project, I'm abandoning the idea of using 10mm diameter motors and "compressible" sponge wheels: there are too many variables on the feasibility and strength of the various components when stressed by the forces generated by the motors on wheels and chassis.

As I mentioned in a previous post, the approach that now seems best to me is to lower the height of the device by employing 6mm diameter motors with 16mm wheels.

To further simplify the design, I could replace the 6mm motors with continuously rotating servos, which should allow me to remove the Motor Driver and thus simplify the wiring.

Many small servos can be found on Hobbyking.com, but their speed is too high for the point I need to explore, well over 100 rpm.

Perhaps you can advise me on a suitable continuous rotation servo ...