FAQ
AI Module
What is the AI Module?
The AI Module is a hardware and software educational tool designed for in-classroom use at the primary and secondary school level. It aims to provide a beginner-friendly and interesting environment where students can learn about artificial intelligence (AI) and development.
In terms of functions, this product possesses machine vision - target detection capabilities. It can recognize user-defined targets and report their position and size.You can use this model to allow robots to perform a wide range of AI functions, such as simple autonomous driving, gesture recognition and following, and automated transportation.
Students will develop AI projects from scratch, directly participating in data collection, labeling, model training, and robot programming. This way, they will grasp the basic theory of AI and gain hands-on experience in AI project development.
In terms of functions, this product possesses machine vision - target detection capabilities. It can recognize user-defined targets and report their position and size.You can use this model to allow robots to perform a wide range of AI functions, such as simple autonomous driving, gesture recognition and following, and automated transportation.
Students will develop AI projects from scratch, directly participating in data collection, labeling, model training, and robot programming. This way, they will grasp the basic theory of AI and gain hands-on experience in AI project development.
Can users use the AI Module to train their own AI models?
Yes.Using the AI Training Platform provided by the DJI Education Hub, users can create their own data sets, train their own AI models, and easily download trained models to hardware modules to run AI recognition algorithms offline.
What robots should I use the AI Module with?
To use the AI Module with the DJI RoboMaster EP (must use the RoboMaster universal expansion module to adapt to the AI module), go to "Create Program" under "My Programs" on the DJI Education Hub for graphical programming, or use the Python SDK for code-based programming. To use the AI Module with the DJI RoboMaster TT, you can use the Python SDK for code-based programming.Both robots have built-in programming interfaces that can be used with the AI Module.For more information, see the Programming Manual and RoboMaster Developer's Manual: https://robomaster-dev.readthedocs.io.
To use the AI Module with a third-party robot platform, you can use the serial port to obtain the identification results returned by the AI Module, but you must parse this data on your own.For more information, refer to AI Module Routines Used By Third-party Controllers.
To use the AI Module with a third-party robot platform, you can use the serial port to obtain the identification results returned by the AI Module, but you must parse this data on your own.For more information, refer to AI Module Routines Used By Third-party Controllers.
How is the AI Module's recognition and reasoning performance?
The built-in model of the AI training platform is deeply optimized for AI computing chips so it can perform recognition and reasoning in only 9ms.At the same time, the AI Module hardware is equipped with a high-speed global shutter camera.It can support most applications requiring motion control.
Do I have to use the AI Module for data collection?
You can use the AI Module to perform data collection. You can also use your own camera device, such as a mobile phone or USB camera, or directly use an online dataset. If you do not have a camera device, you can easily use the image capture mode provided by the AI Module.
How can I increase the recognition frame rate of the AI Module?
Because video streams take up a certain amount of hardware resources and influence the recognition frame rate, you can disconnect the touchscreen from the core module to stop the screen from displaying the video stream. When using the RoboMaster EP robot and Rogram (graphical programming) editor of the DJI Education Hub, you can disable the AI Module video stream to increase the recognition frame rate.
Does the AI Module support the import of other open-source training frameworks or neural network modules?
Because the AI Module uses customized and optimized software and algorithms it does not currently support the import of other open-source training frameworks or the use of different neural network models.
When the AI Module camera sees a plain white wall or desk, the screen displays circular streaks. Is this normal?
This is normal.Because the color depth of the screen is 16 bits, it can only reproduce a limited range of colors. This can produce visible color discontinuity, which appears as streaks on the screen.
After purchasing the hardware, how can I train my own model?
Go to the DJI Education Hub client (downloaded from edu.dji.com), register an account, and obtain an Advanced Edition/Professional Edition license.New users can apply for a 1-month free trial of the Advanced Edition/Professional Edition.After obtaining a license, go to the AI Training Platform of the DJI Education Hub to train your model.Advanced Edition purchase link:https://store.dji.com/cn/product/dji-education-hub-advance-account?vid=101541, Professional Edition purchase link:https://www.dji.com/cn/edu-hub?site=brandsite&from=insite_search#contact
Are the AI Training Platform functions different for the DJI Education Hub trial version, Advanced Edition, and Professional Edition?
Trial version users can only view relevant examples, but cannot create their own tasks. Advanced Edition and Professional Edition users can create a certain number of labeling tasks and algorithm training tasks and publish a certain number of models. The Professional Edition provides a higher task and model limit than the Advanced Edition.In addition, Professional Edition users can purchase an AI Service Package to increase the task and model quota by 3x to 5x and reduce the average algorithm training time. For more information, go to the DJI Education Hub Help Center and refer to Functions for Each DJI Education Hub Version.
Troubleshooting
What should I do if the recognition performance is poor after training a model with the AI Module?
First, make sure your dataset is sufficiently diverse. The following measures can help improve the recognition performance:
1. The dataset should include images of the target from various angles so that the model can properly learn the features of the target.
2. The target should be located in different areas of the images and at different distances from the camera in order to control for lens distortion.
3. The images in the dataset should have environments with diverse textures, colors, and lighting conditions to control for environmental factors.
4. When a single image has multiple instances of a target or multiple targets of different types (such as 2 apples and 1 pear), all the targets must be labeled. Otherwise, unlabeled targets will be considered negative samples, which will affect the training results.
In addition, when creating a new training task, enable the data enhancement and background enhancement options to increase the diversity of the dataset.
Finally, add additional model training steps as appropriate to improve model convergence and achieve better training results.However, more training steps is not always better. Too many steps can significantly increase the training time or lead to overfitting, which reduces the training performance.
1. The dataset should include images of the target from various angles so that the model can properly learn the features of the target.
2. The target should be located in different areas of the images and at different distances from the camera in order to control for lens distortion.
3. The images in the dataset should have environments with diverse textures, colors, and lighting conditions to control for environmental factors.
4. When a single image has multiple instances of a target or multiple targets of different types (such as 2 apples and 1 pear), all the targets must be labeled. Otherwise, unlabeled targets will be considered negative samples, which will affect the training results.
In addition, when creating a new training task, enable the data enhancement and background enhancement options to increase the diversity of the dataset.
Finally, add additional model training steps as appropriate to improve model convergence and achieve better training results.However, more training steps is not always better. Too many steps can significantly increase the training time or lead to overfitting, which reduces the training performance.
When I use the AI Module, the screen does not light up, is all white, or is glitchy. What should I do?
Check the connectors and cable between the core module and touchscreen to make sure they are connected properly. If the screen still does not work, please contact our after-sales service.
There is no response when I insert a microSD card. What should I do?
First, check that the 14-pin extended interface is properly connected to the 14-pin extended interface of the touchscreen module.When using the display screen cable, you cannot use a microSD card because the 14-pin extended interface is not available.
If the device recognizes the microSD card but prompts that it must be formatted, format the SD card after backing up any important data on it.If you still cannot use the card after formatting, try using a different microSD card.
If the device recognizes the microSD card but prompts that it must be formatted, format the SD card after backing up any important data on it.If you still cannot use the card after formatting, try using a different microSD card.
The touchscreen of the AI Module prompts that the microSD card must be formatted. What should I do?
After backing up any important data on the microSD card, tap the option on the touchscreen to format it and try again.If it still does not work, please contact our after-sales tech support.
I cannot properly connect the AI Module with the RoboMaster EP or RoboMaster TT. What should I do?
The RoboMaster EP and RoboMaster TT were adapted to support the AI Module by a firmware update released in May 2022. Please check for updates.
Accessories
When using the AI Module, can I use a third-party Type-C cable instead of the display screen cable?
You cannot use a third-party Type-C cable with the AI Module.The display screen cable used to connect the core module and touchscreen is specially designed for this device. You must use the official cable.
What is the adapter extension board that comes with the AI Module used for?
You can connect the adapter extension board to the 14-pin extension interface of the main board and connect the serial port signal and power lines to the pin header in order to connect a third-party controller to the AI Module using a Dupont cable.The signal names in the silkscreen table below the pin header correspond to the pins of the pin header.For more information, see the AI Module User Manual.
Universal expansion module
What power supply methods are supported by the RoboMaster universal expansion module?
The RoboMaster universal expansion module supports 5V power supply through the micro USB interface and 12V power supply through the CAN interface.
What is the onboard processor chip of the RoboMaster universal expansion module?
The processor of the RoboMaster universal expansion module is the i.MX RT1064DVL6, based on the Cortex-M7 core, with a maximum clock speed of 600 MHz and 4MB on-chip flash and 1MB on-chip SRAM.
How can I update the firmware of the RoboMaster universal expansion module?
Use a CAN data cable to connect the RoboMaster universal expansion module to one of the CAN interfaces on the RoboMaster EP robot and then use the RoboMaster app to perform the update.
How can I use the RoboMaster universal expansion module to obtain images and recognition data from the core module of the AI Module?
Connect the RoboMaster universal expansion module to the core module of the AI module via the 14-pin interface and use a micro USB cable to connect your computer to the universal expansion module.Then, the computer will prompt you to connect a camera device and virtual serial port device. You can use the core module camera to collect images for the computer or you can use the virtual serial port device to communicate with the core module of the AI Module and read recognition data.
Does the RoboMaster universal expansion module support hot swapping?
All interfaces other than the micro USB and CAN BUS interfaces support hot swapping.Forced hot swapping may burn out the onboard chip and irreversibly damage the module.Hot swapping SD cards may corrupt the data on the cards.
Products and services
Does this product come with courses?
Yes, this product comes with a system of courses. Our courses cover computer vision, machine learning, neural networks, and other cutting-edge science and technology topics. In addition, by learning about the RoboMaster EP and AI Module, students will gain a systematic understanding of the technical principles behind AI. This will give them a more complete comprehension of AI.The product also comes with electronic lesson plans and PPTs to help teachers prepare for their classes.To view course information, go to "Learning Resources" on the DJI Education Hub: https://edu.dji.com/hub/teacher/resourceBase?t=536.
How much does the product cost?How can I purchase it?
For price and purchasing inquiries, click "Buy Now" in the upper-right corner and provide your contact information. We will arrange for an expert to contact you as soon as possible.You can also contact your local DJI Education authorized dealer.
What is the age range of the students this product is designed for?
This product is beginner-friendly while providing advanced capabilities and high versatility. It is designed for primary and secondary school students, but university students may also benefit from it.
What sort of teaching venue is this product designed for?
This product can be used for after-school classes for primary and secondary school students, practical learning activities, community courses, science courses, IT courses, general technology courses, and other types of classes. It can even be used for research purposes by university students studying AI and robotics.
What support services are available to individuals, schools, and training institutions?
DJI will provide course materials and instructor training services. For the training schedule, please follow DJI Edu (大疆教育) on WeCom or contact your local DJI Education authorized dealer to get the latest information.