RoboCORE: the heart of your personal robot

[radeknh]

Hi everyone!

I would like to show you the RoboCORE project:
https://www.kickstarter.com/projects/1992634922/robocore-the-heart-of-your-personal-robot



I am the one of the creators from a young startup (our team has 6 persons). I hope that you enjoy RoboCORE
I don’t describe here everything again because you’ll probably visit the KS page anyway.
What I would like to see here are the comments and criticism We want to be opened to the electronics and robotics communities and we will try to answer all your questions.
It would be great for us to see not only the pledges – each comment will help us to create a better final product!

[lukier]

A bit oddly shaped, but I like the design, especially STM32F4.

What motor drivers do you use? Tell me please it's not L293 or SN754410 or anything like that. Lego motors can draw > 2A at peak, some don't have TVS inside, and you need at least overcurrent protection and preferably active rectification.

I'm a TA for a robotics course and since last year we use RPi + an interface board that I won't name, that should never existed in the first place to connect LEGO stuff. Unfortunately, it wasn't my decision to buy this crap (I knew there will be problems when I saw 7805 in a battery powered application  ), the purchase was already made and now I have to repair 5-6 of these boards every week (mostly replacing the H-bridge chips).

I see in the description that you provide some software libraries and MicroPython and even an IDE. Do you have a PID controller in the provided libraries, running in real time on STM32F4 (should be easy, especially with FP support)? This is really useful, especially for students, as by tuning it they can learn a bit about closed loop control (we don't expect them to implement PID).

This year we had to implement a PID in C++ running on RPi side, over UART to the interface board, it works, but it's not perfect (all because interface board is arduino atmega 328 based, duplicated twice - the designers didn't know how to use a bigger MCU and we couldn't be bothered to solder two ISP connectors on 45 boards, fix their buggy arduino fimware and reprogram the whole lot).

If you don't have serious hardware & firmware flaws at every level of the design (like the interface board that I won't name) you have my thumbs up, especially because you have more sensible pricing

[radeknh]

Thank you for the comment.

I wish I could answer every question, but at this moment the hardware is not fully opened. We seriusly consider opening the hardware and this will also depend on the voice of the community. Even if we don't open the hardware, we provide the documentation needed to create custom add-ons. Please have a look on the article here:
http://husarion.com/Hardware_____How_can_I_get_to_it.html

We use DMOS drivers with 1.3A maximum continuous current, 2.5A max peak current and 0.75ohm total resistance. The drivers don't have active rectification function or built-in overcurrent protection, but they are still much better than LB1836 used in LEGO NXT/EV3.
The saturation voltage of L293 or even SN754410 is definitely not acceptable

We provide open libraries for all software which runs on STM32F4, including the PID controller (it uses FPU). PID parameters can be freely modified. The strong point of our system is the hardware quadrature encoder interface for each motor.

[lukier]

I wish I could answer every question, but at this moment the hardware is not fully opened. We seriusly consider opening the hardware and this will also depend on the voice of the community. Even if we don't open the hardware, we provide the documentation needed to create custom add-ons.

No need to be so secretive, especially on this forum, you are not designing a 100 GHz scope. You'll have to open the design at some point, because what if I want to write bare-metal STM32F4 code, without your RTOS/API/MicroPython.

We use DMOS drivers with 1.3A maximum continuous current, 2.5A max peak current and 0.75ohm total resistance. The drivers don't have active rectification function or built-in overcurrent protection, but they are still much better than LB1836 used in LEGO NXT/EV3.

Hmm, not great. Without synchronous rectification you should have freewheel Schottky diodes. Have you considered using DRV8814 or something similar (I think ST has nice motor drivers as well)?

The strong point of our system is the hardware quadrature encoder interface for each motor.

This is the strong point of STM32 timer peripheral and that's why I like them so much for robotic applications.

I wanted to ask what's the interface between Edison and STM32F4? Do you have STM32's ISP UART and BOOT0+RESET pins connected to Edison to allow in system firmware upgrade? If I have 45 of these boards "in the field" I don't want to take them one by one and reflash via SWD or FTDI.

I'm asking these questions because I'm really interested in replacing the current setup for our robotics course (RPi + shameful LEGO interface). The thing is that 150 computer science students (not EEE) are unbelievable testers and the boards operate in very harsh conditions The device needs to have some ESD protection, overcurrent, reverse polarity probably, what would happen if you connected LEGO sensor to a motor port and vice versa, and as I mentioned firmware upgrade in the field (from Edison).

Now, with SN754410 driver that I have to live with now, it is either (depending on how much current the battery can provide) undervoltage, brownout and in effect RPi reset, or the drivers go pop if the battery/power supply can provide > 1.5A.

What I'm saying is that even if I connect your board to my HP server 12V power supply (100A no problem) or a big LiPo pack and toggle the motor direction fast at full speed (huge current spike) or stall the motor nothing should get permanently broken and both Edison and STM32 shouldn't reset.

[radeknh]

No need to be so secretive, especially on this forum, you are not designing a 100 GHz scope. You'll have to open the design at some point, because what if I want to write bare-metal STM32F4 code, without your RTOS/API/MicroPython.

We would like to open the design, but we can't decide alone, as long as we have a financial support from other sides Nevertheless, I also think that opening the hardware is unavoidable.

Hmm, not great. Without synchronous rectification you should have freewheel Schottky diodes. Have you considered using DRV8814 or something similar (I think ST has nice motor drivers as well)?

DRV8814 works only above 8V and it is not acceptable in the device powered from 9V battery pack. The input voltage range was the main problem when we were choosing a driver. For now I don't see a better candidate, maybe we could change the driver to DRV8848... this change would be easy, but we will see. The price is also important.

I wanted to ask what's the interface between Edison and STM32F4? Do you have STM32's ISP UART and BOOT0+RESET pins connected to Edison to allow in system firmware upgrade? If I have 45 of these boards "in the field" I don't want to take them one by one and reflash via SWD or FTDI.

The interface is UART (RX, TX, RTS, CTS) and Edison is able to control BOOT0 and RESET pins of the STM. The ability to reflash the STM via Wi-Fi is one of the most important functionalities. It is also possible to reflash STM via smartphone<->USB<->RoboCORE connection, but it is not an option during a robotic course.

I'm asking these questions because I'm really interested in replacing the current setup for our robotics course (RPi + shameful LEGO interface). The thing is that 150 computer science students (not EEE) are unbelievable testers and the boards operate in very harsh conditions The device needs to have some ESD protection, overcurrent, reverse polarity probably, what would happen if you connected LEGO sensor to a motor port and vice versa, and as I mentioned firmware upgrade in the field (from Edison).

Maybe we should consider creating a very-robust version of RoboCORE for the higher price We are aware that protections are very important, we can only do it slightly better or slightly cheaper. Repairing and replacing damaged devices at our cost is probably more expensive than some additional protections The basic ESD and overcurrent protections or reverse polarity protection are already implemented.

[radeknh]

Hi,

I want to say that our campaign has been succesfully founded
https://www.kickstarter.com/projects/1992634922/robocore-the-heart-of-your-personal-robot

And thank you lukier for your comments! Such practical and specific suggestions are very rare.
Development stage is coming to the end but we will appreciate if anyone post a helpful comment
Thanks!

[radeknh]

I'm writing after a long long time to make an update: we have completed the orders with a few months of delay . We sent the last rewards in March this year.
Nevertheless, the RoboCORE project is now being replaced with Husarion CORE2 - a newer version.
The description is here: https://docs.husarion.com/hardware/core2_1_0_0/index.html

lukier, you were interested in using the RoboCORE for educational purposes. Here you can find one of the best projects that were made at the university which we cooperate with:
https://www.hackster.io/brockhuis-dybi-ski/lego-sweeping-sorting-machine-587968

Thats all, because I don't want to make too much advertising here We are all happy that the campaign was succesful and we constantly need as much feedback as possible.