Forum Replies Created
- Edouard RenardKeymasterSeptember 25, 2019 at 11:30 amPost count: 239
I’ve just made some tests and I can’t reproduce your issue.
System: Windows 10, computer with intel i7, 8-cores, 8GB RAM, tested with Niryo One Studio 2.2.0 (direct download from our website).
I’ve run a sequence with 8 move_joint blocks, inside a loop with 100 iterations –> 800 moves.
- 3 CPUs are used (not 3 threads) with about ~70 threads on 3 CPUs
- Used ressources are: between 10-20% on one CPU, less than 2% on the 2 others
- Memory: About 200-210MB total. When I run the sequence in Blockly (800 moves), the RAM sligthly increases until about 280MB then it drops to 200-210MB again, indefinitely.
- Also I don’t see much lag on the 3D view of Niryo One
Niryo One Studio is based on the Electron framework, which is – I have to admit – a huge resource eater (but it allows an application to be easily integrated and deployed on many OSes). But we’ve made sure that the CPU and RAM usage doesn’t indefinitely increase.
I guess this issue is related to your computer setup, maybe how it handles the Chromium engine (which is the base used by Electron for the app)? Also, please try on another computer to see if you have the same problem.Edouard RenardKeymasterSeptember 25, 2019 at 9:49 amPost count: 239
No, the accuracy of Dynamixel motors stays the same with 2.2.0. As it’s already quite good and works well for educational use, the R&D priority for that is not the highest.
Were you able to make any progress with the motors PID using the debug scripts we previously released?Edouard RenardKeymasterSeptember 25, 2019 at 9:43 amPost count: 239
This certainly is because your robot is quite old, one of the first we sent. Now we don’t have this problem anymore.
If you have an “old” robot (let’s say before 2019), you can add some fans directly on the Raspberry Pi shield (3 fan slots), so you can lower the temperature – standard fan, 12V. Let us know if you have any trouble doing that!Edouard RenardKeymasterSeptember 25, 2019 at 9:39 amPost count: 239Edouard RenardKeymasterSeptember 25, 2019 at 9:30 amPost count: 239Edouard RenardKeymasterSeptember 18, 2019 at 8:44 amPost count: 239Edouard RenardKeymasterSeptember 18, 2019 at 8:41 amPost count: 239Edouard RenardKeymasterSeptember 3, 2019 at 8:47 amPost count: 239Edouard RenardKeymasterAugust 22, 2019 at 10:07 amPost count: 239
There is no better solution, it depends on what you’re trying to do.
Niryo One already has an embedded Raspberry Pi 3, so if you want to use RPi, all you need to do is to connect to it via ssh (tutorial).
If you want to control the robot from the outside, for example to integrate the robot in a complete robotics demonstrators, using an Arduino board is a great idea! This tutorial will help you get started.Edouard RenardKeymasterAugust 22, 2019 at 10:01 amPost count: 239
The end effector tools are not part of the motion planning, here is the part of the robot that is considered to be the TCP (Tool Center Point) for planning
Tools are not included in the robot model, hence not available in Moveit. As for why we chose to do that, here’s a complete discussion on GitHub :)
So if you change tool, yes you might need to add an offset to all commands to get to the same position.Edouard RenardKeymasterAugust 22, 2019 at 9:51 amPost count: 239
In the tutorial provided to remotely execute the Python script, it suggests that I have a Linux OS, isn’t it?
Yes this tutorial is for Linux users, as ssh is not natively supported on Windows (yet?), the command won’t work.
So adding a new node like you proposed seems a good idea.
About the delay, do you mean something like “time.sleep(20)” at the beginning of the main function?
Yes it should work. If you have some problems when testing don’t hesitate to increase the delay.Edouard RenardKeymasterAugust 20, 2019 at 9:20 amPost count: 239
Which file shows me the dynamixel and can configuration for the version Niryo One Raspberry Pi 3 image v 1.1.0
dxl baudrate and dxl uart device name ? and can spi channel , spi baudrate and gpio can interrupt ?
Ah ok, thanks for clarifying I think I understand what you mean:
So, first of all, there is the hardware version (V1, before September 2018 and V2, after September 2018). Then there is the software version, starting at 1.0.0, now 2.1.0, and soon a new upgrade. Whether you have hardware V1 or V2 is the same, as software versions are handling both V1 and V2.
The config you’re looking for is in the niryo_one_driver.yaml file, from software version 1.1.0 until now. It hasn’t changed, so you already have found the correct values.
Note that using software v1.1.0 is not recommended, since then we have added lots of new functionalities and fixed many bugs. Please update to the newest version :)Edouard RenardKeymasterAugust 20, 2019 at 9:10 amPost count: 239
There is a better way to do what you want to do. You can simply remotely execute the Python script from your own computer. We have made a tutorial for that here.
If however, you plan to add a node into an existing launch file, I suggest you put it in user_interface.launch, at the end. And inside the node, add a delay of about 15-20 seconds so that all the other functionalities are correctly started.Edouard RenardKeymasterAugust 14, 2019 at 6:15 pmPost count: 239Edouard RenardKeymasterAugust 9, 2019 at 2:05 pmPost count: 239
On your linux machine you should setup a manual network, with the network being 169.254.0.0 (netmask 16) and the gateway being 169.254.1.1.
On the Raspberry Pi the IP is static: 169.254.200.200, with the gateway 169.254.1.1
There is currently an issue on the current Raspberry Pi image (v 2.1.0). There is a route missing, so the Ethernet connection is currently broken. For now you can add it by yourself to make it work. This issue will be fixed on the next software release (beginning of September 2019).