ftc265 is a plug-and-play Intel RealSense T265 VSLAM camera wrapper for FTC based off my work in FRC writing a roboRIO T265 wrapper.
To get started you can either copy the example project, or you can install manually.
Paste the following into TeamCode/build.gradle:
repositories {
// This line will already be here; you do not need to add it
maven { url = "https://dl.bintray.com/first-tech-challenge/ftcsdk/" }
// Add this line
maven { url = "https://maven.0x778.tk" }
}At the bottom of TeamCode/build.gradle add the following (if you have an existing dependencies block you can just add the implementation line to that dependencies block):
dependencies {
// There should be a bunch of implementation configurations here already
implementation 'com.spartronics4915.lib:ftc265:2.1.9'
}Finally, set the sourceCompatibility, targetCompatibility, and minSdkVersion in the build.common.gradle at the project root if you haven't already:
android {
// NOTE: There will be *way* more things in this file than what's shown here.
// The blocks shown here should already exist, and you'll only need to change these three shown parameters.
defaultConfig {
minSdkVersion 24
}
compileOptions {
sourceCompatibility JavaVersion.VERSION_1_8
targetCompatibility JavaVersion.VERSION_1_8
}
}You'll need to perform a Gradle sync in Android studio (if you use that IDE) after adding the new repositories and dependencies.
Basic usage is as follows:
// This is the transformation between the center of the camera and the center of the robot
Transform2d cameraToRobot = new Transform2d();
// Increase this value to trust encoder odometry less when fusing encoder measurements with VSLAM
double encoderMeasurementCovariance = 0.8;
// Set to the starting pose of the robot
Pose2d startingPose = new Pose2d(1, 1, new Rotation2d());
T265Camera slamra = new T265Camera(cameraToRobot, encoderMeasurementCovariance, hardwareMap.appContext);
slamra.setPose(startingPose); // Useful if your robot doesn't start at the field-relative origin
// Call this when you're ready to get camera updates
slamra.start();
// Now we can grab our last received pose in our main thread
while (true) {
slamra.getLastReceivedCameraUpdate();
}There is also a ready-to-use example project here.
Please note that the above example uses the simple synchronous API. There is also a more advanced callback-based API available.
If you're getting a "No camera connected" exception then you can try a few things:
- Connect the camera to a computer running realsense-viewer and make sure that it works there.
- If you're using a USB hub, make sure to plug the hub in first, power the hub on (if applicable), and then connect the camera.
- If you're using the Control Hub and your camera won't connect when you first boot up, try plugging the camera into the USB 2 port on subsequent boots.
- If nothing works then you can open an issue or join the Discord server.
This is likely happening because you haven't granted the correct permissions. The easiest way to get a permissions prompt is to uninstall and reinstall the robot controller app. After reinstalling you should get a permissions prompt on first run.
It's possible that you used the wrong overload of start. If you used the overload of the start method that takes a Consumer<CameraUpdate> (a camera update callback) then you won't be able to use getLastRecievedCameraUpdate unless you use the overload of start that takes no parameters.
According to this GDC clarification, yes.
If that doesn't convince you, please take the following into account:
- In the opinion of the author, this is no more of a coprocessor than a servo. The T265 has an ASIC just like a servo might have a microcontroller, but critically, neither of these are reprogrammable. Similar devices that do offboard nonprogrammable video processing have been allowed (e.g. the Pixy cam, which was cited in the GDC clarification linked above.)
- This fits under the broad category of UVC-compatible devices.
In my testing, yes. I used this in-season at (more demanding) FRC speeds, and when testing on an FTC-field-sized area I consistiently got less than 1 inch of error.
- The T265 is smaller and easier to mount and maintain than sprung odometry wheels. It is also likely as accurate.
- Awards.
Yes. This should work out of the box on the Control Hub.
Yes.
Unfortunately, there is no canonical set of geometry classes (Pose2d, Transform2d, etc.) in FTC so I just had to pick a set to use. I prefer the FTCLib geometry classes because they're directly copied from WPILib (which I'm quite familiar with.) I also think that these WPILib-style geometry classes are better than the alternatives.
I had to write JNI bindings to C++ librealsense because no Java bindings exist for the T265. Then I had to patch librealsense to get around some bugs and Android incompatibilities (especially related to the maximum size for USB bulk transfers and the list of USB VID/PID combos); this is why I distribute a custom version of librealsense.
Yes. Please report bugs or (even better) make a PR. FTC is not my main focus, and I do not currently have access to testing hardware, so I would appreciate it if any nontrivial PRs are accompanied by a test on real hardware (or even better, a unit test.)
This is contingent on the community's response and Intel's actions. If ftc265 becomes popular then I will continue to support it as best I can into the future until Intel provides a reasonable alternative. Please note that I have limited time and that FTC is not my main focus; I will endevor to fix issues and respond to you as fast as possible, but expect the standard deviation on my response time to be high :)
Copyright 2020 Declan Freeman-Gleason
This program is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU Lesser General Public License along with this program. If not, see https://www.gnu.org/licenses/.