FTCLib: A Java repository from micahreich

FTCLib

origin: https://github.com/FTCLib/FTCLib

upstream: https://github.com/OpenFTC/OpenRC-Turbo

Contributors: Please read CONTRIBUTING.md

OpenRC is a modified version of the official FTC SDK in which all of the source code that is normally tucked away inside the AAR files has been extracted into modules. This makes it easy to see and modify almost the entirety of the Robot Controller app's source code. In addition, the history in Git shows all changes that have been made to the core code since OpenRC's inception. This complements the changelogs that FIRST provides, allowing teams to see exactly what code has been changed.

Legality for competition use

According to the 2019-2020 Game Manual Part 1, teams are not allowed to replace or modify the portions of the SDK which are distributed as AAR files, per <RS09>. This means that in its default configuration, OpenRC is not legal for competition.

HOWEVER, in order to address this, OpenRC has a stock build variant which will compile the TeamCode and FtcRobotController modules against the official, unmodified AAR files, rather than against the extracted modules.

Device compatibility

Unfortunately, OpenRC is only compatible with devices that run Android 6.0 or higher. For FTC, this means that it is incompatible with the ZTE Speed. OpenRC will work fine on all other FTC-legal devices (including the new Control Hub).

For the curious: the cause of the incompatibility is the result of a bug in the dlopen() function of Android versions prior to 6.0. When loading the libRobotCore.so on older Android versions, an UnsatisfiedLinkError will be thrown because it cannot find a symbol that is declared in libVuforia.so and dlopen() is not smart enough to know that libVuforia.so has already been loaded into memory. See the "Correct soname/path handling" section of this page for more details.

Build variants

Variant Descriptions

Stock - 40MB APK (oof!)
- Competition legal
Turbo - 10MB APK (4x smaller!)

Note: If you would like to use Blocks, you will need to copy your private Vuforia key into the Blocks/src/main/assets/CzechWolf file
- Vuforia native library loaded dynamically
- Vuforia/TF datasets loaded dynamically
- OnBotJava removed
Extreme Turbo - 4MB APK (10x smaller!)
- Vuforia native library loaded dynamically
- Vuforia/TF datasets loaded dynamically
- OnBotJava removed
- Blocks removed
- Web management removed
- Sound files removed

Benchmarks

	Nexus 5 (7.1.2) Intel Centrino Advanced-N 6205 on Linux	Nexus 5 (7.1.2) Intel Wireless 8260 on Linux
Stock over WiFi	30 sec	20 sec
Turbo over WiFi	13 sec	11 sec
Extreme Turbo over WiFi	10 sec	8 sec

Setup Process

Fork this repository
Clone your fork
Do git remote add upstream https://github.com/OpenFTC/OpenRC-Turbo.git
Copy all of the files found in the filesForDynamicLoad folder of this repo into the FIRST folder on the RC's internal storage
Select your desired build variant (see the Switching Build Variants section)

Update Process

Assuming you followed the above setup process, all that you need to do to update your fork when a new OpenRC release is available is:

git pull upstream master
Perform a Gradle Sync
If the project fails to build, try Clean Project, Rebuild Project, and Invalidate Caches / Restart

Switching Build Variants

IMPORTANT: make sure to test that your project compiles correctly with the stock variant at least a week before your competition!

Note: you may get a "variant conflict" when switching variants. You can fix this by changing the conflicting module's variant to match the variant you want.

Open the Build Variants tab in the lower left hand corner of Android Studio
In the dropdown for the TeamCode module, select your desired variant
Perform a Gradle sync

Versioning Scheme

To prevent confusion, OpenRC does not have its own version number. The version number will directly reflect the SDK version that the release is based on. However, the version number will have a letter appended to the end of it, which will be incremented (A-Z) for each release of OpenRC which is based on the same SDK version. When OpenRC is updated to be based on a new SDK version, the letter will reset to A.

For instance, the 3rd release of OpenRC based on SDK v5.0 would be 5.0C, whereas the first release of OpenRC based on SDK v5.1 would be 5.1A.

Dynamic Loading of TensorFlow and Vuforia Datasets

In order to reduce APK size, the Turbo and Extreme Turbo variants do not bundle the Vuforia and TensorFlow datasets in the APK. However, once copied onto the internal storage (see step #4 in the Setup Process section), you can still use them in your OpMode by making a very minor change.

NOTE: The samples in this repo have already been adjusted with this change.

/*
 * For Vuforia
 */
// Find a line similar to this in the program
vuforia.loadTrackablesFromAsset("DatasetName");
// And replace it with this (obviously adjusting the DatasetName)
vuforia.loadTrackablesFromFile("/sdcard/FIRST/DatasetName");

/*
 * For TensorFlow
 */
// Find a line similar to this in the program
tfod.loadModelFromAsset("DatasetName.tflite");
// And replace it with this (obviously adjusting the DatasetName)
tfod.loadModelFromFile("/sdcard/FIRST/DatasetName.tflite");

Release Notes:

5.3B

Released on 22 November 2019

Fix TFOD crash on stock due to incorrect version of TFOD library being used (which conflicted with pre-compiled official FTC SDK AARs)

5.3A

Released on 19 October 2019

Update to SDK v5.3

5.2B

Released on 19 September 2019

Remove universal Blocks Vuforia key per request of FIRST. If you would like to use Blocks, you will need to copy your private Vuforia key into the Blocks/src/main/assets/CzechWolf file.

5.2A

Released on 11 September 2019

Update to SDK v5.2
NOTE: You will need to copy some additional files to the FIRST folder of the internal storage after you update. See step #4 in the Setup Process section
NOTE: The TensorFlow and Vuforia sample OpModes for SKYSTONE have been modified slightly to load the datasets from internal storage. Use the samples in this repo; the stock samples will fail on variants other than stock. Please also see the Dynamic Loading of TensorFlow and Vuforia Datasets section of this readme.

5.1A

Released on 26 August 2019

Update to SDK v5.1
Updated dynamic Vuforia loader to enforce being run on Android 6.0 or higher

5.0A

Released on 21 August 2019

Initial release.

micahreich/FTCLib