4WD with 2 encoders only?

[Pito Salas]

I am designing a new robot. My plan is to make it 4WD and initially skid steering, to see how it goes.

My intuition says that all four motors should have encoders so that I can control the RPM for steering etc.

I have read in more than one place that an option is to have the front and back motors on each side be powered together. In other words, the rear left wheel will have a motor and encoder. The front left wheel will not have an encoder. The proposition is that you power both left wheels in parallel (given the right motor controller) and use the single encoder to control the RPM. The theory is that given the same power and identical motors the front and back wheels will spin at “approximately” the same speed. I’ve seen a claim that it is “standard practice”.

Remember this is not a track and the two wheels are not chained together or have any mechanical mechanism to keep them in sync.

Have you seen this and does it actually work?

Thanks!

Pito

Boston Robot Hackers &&
Comp. Sci Faculty, Brandeis University (Emeritus)

[Karim Virani]

"initially skid steering" is interesting by what it doesn't say. 

i use a p3at skid steer chassis with 4 wheels, 4 motors, 2 of which have encoders on them. but my system is mechanically linked - a belt gangs the leader (encodered) and follower motors internally. this is a common configuration for a skid steer in the mid 2000s. with either treads or transmission ganging one side together, you can be fairly confident the encoder is giving you the average motion on that side.

yes you can pair motors by their tendency to match performance under load, and then just give the unlinked follower the same power levels. you will be less certain about your average motion. note that this can be an advantage or disadvantage on certain terrain types. what it does mean, particularly with a suspensionless design - is that a non-encoded motor will often surge when it loses contact with the ground. when it re-engages, there can be a twist impulse. but perhaps you are intending some kind of suspension which is also why you aren't mechanically ganging the front and rear motors?

that said, it can definitely work. but these days i'd recommend 4 encoders and also current / power sense on each motor. so you can be more sure about what is happening under stall / near-stall conditions.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/B96A81C5-E360-457B-AB54-EADDB1C49E6A%40brandeis.edu.

[Ken Gregson]

That's what you are typically "stuck" with in most inexpensive off the shelf 4WD robot bases. Not sure how to ensure both wheels on the same side are actually turning at the same speed without an encoder on all wheels though (you know what they say about assumptions). And there will be slip no matter what.

Found this project using 4:

https://hackaday.io/project/25406-wild-thumper-based-ros-robot/log/61787-2wd4wd-wheel-odometry

Still needs an IMU, 2 might be enough?

--
Check out https://bostonrobothackers.com!
---
You received this message because you are subscribed to the Google Groups "Boston Robot Hackers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to boston-robot-hac...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/boston-robot-hackers/B96A81C5-E360-457B-AB54-EADDB1C49E6A%40brandeis.edu.

[Chris Albertson]

Wire both the left side motoer in parallel and the right side also in parallel.  Now in effect you have a two-motor chassis and you only need a left and right encoder.

Two moters in parallel wil run very close to the same speed.  So you get by with one moter controller H-bridge

I did this with a 6-wheel drive chassis and only use encoders in the center pair of weeks and treated the robot as a two-wheel diferential drive.

One more thing.  Have you seen magnetic encoders?  They cost almost nothing and are the size of postage stamps.   but they need to “see” the end of the shaft https://www.amazon.com/dp/B094F8H591/ref=vp_d_fuw_pd?th=1.  They are good enough to put on wheels, almost good enough for motor shafts.

-- 
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/CAArUYGtYNLv8DhpoSdzec3YBj0Xb8uNPchAT%3DySR_fbJjmUFCA%40mail.gmail.com.

[Pito Salas]

Chris, and all, thanks. Just to be clear that the front and back motors are not mechanically “ganged” together with a chain, belt or track. They are totally separate.

In your example, how well did odometry from the encoders work?

Best,

Pito

Boston Robot Hackers &&
Comp. Sci Faculty, Brandeis University (Emeritus)

> On Jun 1, 2026, at 3:03 PM, Chris Albertson <alberts...@gmail.com> wrote:
>
> Wire both the left side motoer in parallel and the right side also in parallel. Now in effect you have a two-motor chassis and you only need a left and right encoder.
>
> Two moters in parallel wil run very close to the same speed. So you get by with one moter controller H-bridge
>
> I did this with a 6-wheel drive chassis and only use encoders in the center pair of weeks and treated the robot as a two-wheel diferential drive.
>

> One more thing. Have you seen magnetic encoders? They cost almost nothing and are the size of postage stamps. but they need to “see” the end of the shaft https://www.amazon.com/dp/B094F8H591/ref=vp_d_fuw_pd?th=1 [amazon.com]. They are good enough to put on wheels, almost good enough for motor shafts.

>
>
>
>> On Jun 1, 2026, at 9:57 AM, Ken Gregson <ken.g...@gmail.com> wrote:
>>
>> That's what you are typically "stuck" with in most inexpensive off the shelf 4WD robot bases. Not sure how to ensure both wheels on the same side are actually turning at the same speed without an encoder on all wheels though (you know what they say about assumptions). And there will be slip no matter what.
>>
>> Found this project using 4:

>> https://hackaday.io/project/25406-wild-thumper-based-ros-robot/log/61787-2wd4wd-wheel-odometry [hackaday.io]

>>
>> Still needs an IMU, 2 might be enough?
>>
>> On Mon, Jun 1, 2026, 9:43 AM Pito Salas <pito...@gmail.com> wrote:
>> I am designing a new robot. My plan is to make it 4WD and initially skid steering, to see how it goes.
>>
>> My intuition says that all four motors should have encoders so that I can control the RPM for steering etc.
>>
>> I have read in more than one place that an option is to have the front and back motors on each side be powered together. In other words, the rear left wheel will have a motor and encoder. The front left wheel will not have an encoder. The proposition is that you power both left wheels in parallel (given the right motor controller) and use the single encoder to control the RPM. The theory is that given the same power and identical motors the front and back wheels will spin at “approximately” the same speed. I’ve seen a claim that it is “standard practice”.
>>
>> Remember this is not a track and the two wheels are not chained together or have any mechanical mechanism to keep them in sync.
>>
>> Have you seen this and does it actually work?
>>
>> Thanks!
>>
>> Pito
>>
>> Boston Robot Hackers &&
>> Comp. Sci Faculty, Brandeis University (Emeritus)
>>
>> --

>> Check out https://bostonrobothackers.com [bostonrobothackers.com]!

>> ---
>> You received this message because you are subscribed to the Google Groups "Boston Robot Hackers" group.
>> To unsubscribe from this group and stop receiving emails from it, send an email to boston-robot-hac...@googlegroups.com.

>> To view this discussion visit https://groups.google.com/d/msgid/boston-robot-hackers/B96A81C5-E360-457B-AB54-EADDB1C49E6A%40brandeis.edu [groups.google.com].

>>
>> --
>> You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
>> To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

>> To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/CAArUYGtYNLv8DhpoSdzec3YBj0Xb8uNPchAT%3DySR_fbJjmUFCA%40mail.gmail.com [groups.google.com].

>
>
>
> --
> You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

> To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/37573468-E8CB-4C57-9DCE-0D9C1E6CB2F0%40gmail.com [groups.google.com].

[Chris Albertson]

> On Jun 1, 2026, at 12:05 PM, Pito Salas <pito...@gmail.com> wrote:
>
> Chris, and all, thanks. Just to be clear that the front and back motors are not mechanically “ganged” together with a chain, belt or track. They are totally separate.
>
> In your example, how well did odometry from the encoders work?

Mine was a 6WD chasis with encoder only on the center pair. For outdoor use odometry is nearly usless, big holes, driving over bricks and rocks. The encoders were for speed control, not so much navigation.

I think the key for doing navigation with odometry is to prevent wheels from slipping. With a 4 wheel chassis, you need suspention if the floor is not perfect or else you will have the weight on only the two diagonal wheels.

Later work I am using those magnetic encodrs or actualy some better quality 14-bit versions. They are easy to use bacasue thay are not quadrature but direct reading and you query them over SPI interface. Place on on the motor and one after the gear reduction

> To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/B72B1D01-BEAD-4F64-840E-18FF0FB3ECD4%40brandeis.edu.

[Pito Salas]

Have you seen this device, looks interesting…anyone have experience with it?

Best,

Pito Salas

Boston Robot Hackers &&

Computer Science Faculty, Brandeis University

On Jun 1, 2026, at 3:46 PM, Chris Albertson <alberts...@gmail.com> wrote:



To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/9EAF8473-B707-418A-BA46-79DB16A69E3B%40gmail.com.

[Karim Virani]

my FTC teams have used it. not reliable  enough for me. sensitive to the kind of floor. not suitable for outdoors. not useable for higher speed robots, but can do approximately walking speed.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/95857DFD-2E91-42D7-82EC-D22D65CF4402%40brandeis.edu.

[Chris Albertson]

I don’t think you can depend on any one device for perfect navigation.   Wheels will slip, gyros will drift.   I think you have to synthesise from multiple imperfect sources.

[Ken Gregson]

Sorry, realized this seems to be going to two separate groups at the same time?

So, the question really depends on what you need and are trying to achieve and with what processing power.

The performance of the original Neato XV-11 for it's navigation accuracy and precision (indoors) using a (very clever) single beam lidar and odometery blew me away. First/best mapping, planning, and navigation to vacuum the floor with the most space/energy efficient lawnmower pattern. Immediately bought one when I saw that, sad they are gone.

The SEN-24904 quotes 3-5% accuracy, maybe reducible to 1% with (careful) calibration - not great. Maybe Franklin (BRH) has data from an XRP (for which it was designed)?

This paper has data with analysis that suggests single wheel odometer + gyro (cutely referred to as OG) could get down to 0.26% (translation error) with quality sensors and careful calibration - in ideal conditions with no wheel slip. But degrades significantly when those conditions aren't met. https://arxiv.org/pdf/2505.04438

Agree there is no one "silver bullet" sensor. Lidar, radar, depth, vision, accelerometer, gyro, mag, GPS could all contribute and need to be fused for good robust state estimation. If you have the processing power (and budget) for it all?

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/486D807E-BA25-42E7-B15B-8D6F5E6FFF1D%40gmail.com.