Advices for the SPOTMICRO project

[Raph1821]

Dear Mr. Mike4192,

I am reaching out to kindly request your guidance regarding the SPOTMICRO project. My name is Minh Nhut NGUYEN, and I am currently completing the final year of my Master’s degree in Automatic Control and Robotics, within the Intelligent Systems Engineering program at Sorbonne University in Paris, France.
For my Master’s Thesis Project, I have chosen to work on your SPOTMICRO project, and I am leading a team responsible for building and further developing the robot.

We have made several adaptations to ensure compatibility with a Raspberry Pi 4B (8 GB) running Ubuntu 20.04 and ROS1 Noetic, which we intend to use on the final robot. At this stage, we have successfully tested your original program and simulation. You may take a look at our repository for more details:
https://github.com/Raph1821/SPOTMICRO__PFE__2025__SORBONNE__SACLAY

We are currently encountering some difficulties during the physical construction of the robot, particularly concerning the plate design. Although the GitHub repository provides details regarding the use of a LiPo battery, we were unable to find pictures or schematics indicating its exact placement. Additionally, we plan to integrate an IMU sensor, which would require further modifications to the plate.

If you would be willing to advise us, or simply exchange a few technical insights regarding these points, it would be extremely helpful.

[mike4192]

Hi Minh, I'm glad to hear you have been able to adapt and work with my project and get it running on a Rpi 4 and on Ubuntu 20.04. The physical design and construction of this robot is definitely a weak point of the robot platform overall. I myself did not design it but merely used and 3d printed a design that existed at the time, adapting it as needed to get a working result. You will likely find the assembly is tedious and annoying as parts need to be assembled in a certain order of operations. The integration of servos into the legs adds a lot of mass to the legs themselves, which puts strain on the various mechanical joints and limits overall walking performance, and increases wear and tear and slop to develop. Most state of the art quadrupeds use brushless motors and command torque at the lowest level (rather than position via servos), and use belts to distribute forces to leg joints. But alas, that is more complex and expensive. In regard to battery placement, from an older video <https://photos.app.goo.gl/5DPK4c5SzDjycH4EA> I was able to dig up, it looks like I crudely zip-tied a lipo battery to the front of the robot. It looks like I was using a 2 cell 4000 mAh lipo. In general I believe I positioned the battery so as to position the center of mass of the robot to be approximately centered between the 4 legs, as that provided for the best walking performance. For mounting an IMU, I would also suggest positioning it as close as possible to the robot's center of mass, via some sort of custom 3d printed bracket or mounting adapter and glueing or screwing it into the robot's frame. Filtering the IMU data to obtain good enough feedback signals for closed loop control will definitely pose a challenge.

…

-Mike

On Thu, Nov 20, 2025 at 9:20 AM Minh Nhut NGUYEN ***@***.***> wrote: *Raph1821* created an issue (mike4192/spotMicro#96) <#96> Dear Mr. Mike4192, I am reaching out to kindly request your guidance regarding the SPOTMICRO project. My name is Minh Nhut NGUYEN, and I am currently completing the final year of my Master’s degree in Automatic Control and Robotics, within the Intelligent Systems Engineering program at Sorbonne University in Paris, France. For my Master’s Thesis Project, I have chosen to work on your SPOTMICRO project, and I am leading a team responsible for building and further developing the robot. We have made several adaptations to ensure compatibility with a Raspberry Pi 4B (8 GB) running Ubuntu 20.04 and ROS1 Noetic, which we intend to use on the final robot. At this stage, we have successfully tested your original program and simulation. You may take a look at our repository for more details: https://github.com/Raph1821/SPOTMICRO__PFE__2025__SORBONNE__SACLAY We are currently encountering some difficulties during the physical construction of the robot, particularly concerning the plate design. Although the GitHub repository provides details regarding the use of a LiPo battery, we were unable to find pictures or schematics indicating its exact placement. Additionally, we plan to integrate an IMU sensor, which would require further modifications to the plate. If you would be willing to advise us, or simply exchange a few technical insights regarding these points, it would be extremely helpful. You may reach me using any of the contact details below : University Mail : ***@***.*** Gmail : ***@***.*** Discord : minh_nhut_n What's app : 0773801812 — Reply to this email directly, view it on GitHub <#96>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALYO2GNTMPKJLJRVKO7BIT35XZ6HAVCNFSM6AAAAACMW7545WVHI2DSMVQWIX3LMV43ASLTON2WKOZTGY2DQMRYGQ2DEOI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Raph1821]

Hi Mike,

Thank you for your response. I will try to find a suitable solution for modifying the plate to accommodate the battery placement, and I will keep you informed once we manage to resolve this issue.

In addition, I would like to ask whether a circuit diagram exists for your version of the robot. If such a diagram is available, it would be extremely helpful to us, as we have found many different versions online. At the moment, the only assembly references we can rely on are KDY’s videos, but his design uses an Arduino board, which differs from your implementation.

Once again, thank you for your assistance.

Minh Nhut

[mike4192]

Hi Minh, If you haven't already seen them, check out the photos under the "Sample Hardware Install Photos" section at the bottom of the addition_hardware_description_page <https://github.com/mike4192/spotMicro/blob/master/docs/additional_hardware_description.md> which shows placement of the RPi and PCA9685 board. I never created a proper circuit diagram but drew the following from memory which I think captures the major elements. Note that the servos were directly connected to the battery (not through a voltage regulator) so they could receive maximum current as servos can draw a lot of current during high torque operation. The PDI-HV5523MG servos are considered "high voltage" servos in that they support a 6-8.4V input voltage, whereas regular servos typically have a 4.6-6V operating range (which would be fried if directly connected to the 7.4 V 2s lipo). The UBEC (which is really just a 5v regulator) was used to power the Raspberry Pi and PCA9685. I used a "UBEC" which is a voltage regulator from RC plane applications as it was what I had readily available. Note that raspberry pis need considerable power supply capacity, so make sure whatever voltage regulator you use can supply at least 3+ amps. The PC9685 itself has negligible power draw, and is connected to the Pi via i2c 2 wire connection.

…

On Fri, Nov 21, 2025 at 5:34 AM Minh Nhut NGUYEN ***@***.***> wrote: *Raph1821* left a comment (mike4192/spotMicro#96) <#96 (comment)> Hi Mike, Thank you for your response. I will try to find a suitable solution for modifying the plate to accommodate the battery placement, and I will keep you informed once we manage to resolve this issue. In addition, I would like to ask whether a circuit diagram exists for your version of the robot. If such a diagram is available, it would be extremely helpful to us, as we have found many different versions online. At the moment, the only assembly references we can rely on are KDY’s videos, but his design uses an Arduino board, which differs from your implementation. Once again, thank you for your assistance. Minh Nhut — Reply to this email directly, view it on GitHub <#96 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AALYO2GTSRWRNVN5ADGS333354IFJAVCNFSM6AAAAACMW7545WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTKNRTGA2TMOBUGE> . You are receiving this because you commented.Message ID: ***@***.***>

[mike4192]

I think my diagram didn't get attached by email, so here it is again.

[chrisalbertson]

I did a redesign. of SM to try and fix what I thought was a weak shoulder design and some other problems. I also wanted more room for battery and electronics. This worked. And I think the Fusion360 CAD files are on my Github page. The problem with SM is that iit uses hobby servos. Even the best higher-end servos lack the performance you want for a dog-robot. It woiuld be MUCH better to use BLCD drone moters with about 9:1 reduction. But thjis raises the cost to well over $1,000. What I did that I think improves the life of the servoes is suported the weight with ball bearing units and used the motors onlyu to provide torque to the joints. The orginial design put bending loads on ther servo output shafts and would quickly wear out the output shaft bearings Today, companies like Unitree sell a working robot dog for less then I can buy the parts for. I disagree with the advice below to mount the IMU st the robot’s center of mass. This would make some transform simpler but if you are using ROS Localization, and ROS’ Kalman filter. The very best practice is to use two IMU units (they are cheap) and place one at each end. This way the accelerometers can help with the rotation and ros’ TF and KR do the math for you.

…

On Nov 20, 2025, at 10:31 AM, mike4192 ***@***.***> wrote: mike4192 left a comment (mike4192/spotMicro#96) <#96 (comment)>Hi Minh, I'm glad to hear you have been able to adapt and work with my project and get it running on a Rpi 4 and on Ubuntu 20.04. The physical design and construction of this robot is definitely a weak point of the robot platform overall. I myself did not design it but merely used and 3d printed a design that existed at the time, adapting it as needed to get a working result. You will likely find the assembly is tedious and annoying as parts need to be assembled in a certain order of operations. The integration of servos into the legs adds a lot of mass to the legs themselves, which puts strain on the various mechanical joints and limits overall walking performance, and increases wear and tear and slop to develop. Most state of the art quadrupeds use brushless motors and command torque at the lowest level (rather than position via servos), and use belts to distribute forces to leg joints. But alas, that is more complex and expensive. In regard to battery placement, from an older video <https://photos.app.goo.gl/5DPK4c5SzDjycH4EA> I was able to dig up, it looks like I crudely zip-tied a lipo battery to the front of the robot. It looks like I was using a 2 cell 4000 mAh lipo. In general I believe I positioned the battery so as to position the center of mass of the robot to be approximately centered between the 4 legs, as that provided for the best walking performance. For mounting an IMU, I would also suggest positioning it as close as possible to the robot's center of mass, via some sort of custom 3d printed bracket or mounting adapter and glueing or screwing it into the robot's frame. Filtering the IMU data to obtain good enough feedback signals for closed loop control will definitely pose a challenge. -Mike On Thu, Nov 20, 2025 at 9:20 AM Minh Nhut NGUYEN ***@***.***> wrote: > *Raph1821* created an issue (mike4192/spotMicro#96) > <#96> > > Dear Mr. Mike4192, > > I am reaching out to kindly request your guidance regarding the SPOTMICRO > project. My name is Minh Nhut NGUYEN, and I am currently completing the > final year of my Master’s degree in Automatic Control and Robotics, within > the Intelligent Systems Engineering program at Sorbonne University in > Paris, France. > For my Master’s Thesis Project, I have chosen to work on your SPOTMICRO > project, and I am leading a team responsible for building and further > developing the robot. > > We have made several adaptations to ensure compatibility with a Raspberry > Pi 4B (8 GB) running Ubuntu 20.04 and ROS1 Noetic, which we intend to use > on the final robot. At this stage, we have successfully tested your > original program and simulation. You may take a look at our repository for > more details: > https://github.com/Raph1821/SPOTMICRO__PFE__2025__SORBONNE__SACLAY > > We are currently encountering some difficulties during the physical > construction of the robot, particularly concerning the plate design. > Although the GitHub repository provides details regarding the use of a LiPo > battery, we were unable to find pictures or schematics indicating its exact > placement. Additionally, we plan to integrate an IMU sensor, which would > require further modifications to the plate. > > If you would be willing to advise us, or simply exchange a few technical > insights regarding these points, it would be extremely helpful. > You may reach me using any of the contact details below : > > University Mail : ***@***.*** > Gmail : ***@***.*** > Discord : minh_nhut_n > What's app : 0773801812 > > — > Reply to this email directly, view it on GitHub > <#96>, or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AALYO2GNTMPKJLJRVKO7BIT35XZ6HAVCNFSM6AAAAACMW7545WVHI2DSMVQWIX3LMV43ASLTON2WKOZTGY2DQMRYGQ2DEOI> > . > You are receiving this because you are subscribed to this thread.Message > ID: ***@***.***> > — Reply to this email directly, view it on GitHub <#96 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABQKNRR46O5IZOQE4IJLFJL35YCHTAVCNFSM6AAAAACMW7545WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTKNJZGQ3TGMZYGY>. You are receiving this because you are subscribed to this thread.

[Raph1821]

Thank you guys for all of the helpful informations. I will try them and keep you guys informed once we succeed.