sandilyasg
commented
5 months ago Example script below
// Copyright (c) Microsoft Corporation. All rights reserved.
// Licensed under the MIT License.
#include "common/common_utils/StrictMode.hpp"
STRICT_MODE_OFF
#ifndef RPCLIB_MSGPACK
#define RPCLIB_MSGPACK clmdep_msgpack
#endif // !RPCLIB_MSGPACK
#include "rpc/rpc_error.h"
STRICT_MODE_ON
#include "vehicles/multirotor/api/MultirotorRpcLibClient.hpp"
#include "common/common_utils/FileSystem.hpp"
#include <iostream>
#include <chrono>
int main()
{
using namespace msr::airlib;
msr::airlib::MultirotorRpcLibClient client;
typedef ImageCaptureBase::ImageRequest ImageRequest;
typedef ImageCaptureBase::ImageResponse ImageResponse;
typedef ImageCaptureBase::ImageType ImageType;
typedef common_utils::FileSystem FileSystem;
try {
client.confirmConnection();
std::string drone_name;
std::cout << "Enter the name of quadrotor e.g. Drone1" << std::endl;
std::cin >> drone_name;
std::cout << "Press Enter to get FPV image from " << drone_name << std::endl;
std::cin.get();
// const std::vector<ImageRequest> request{ ImageRequest("0", ImageType::Scene), ImageRequest("1", ImageType::DepthPlanar, true), ImageRequest("2", ImageType::Segmentation) };
// check how the segmentation classes can be obtained here
// const std::vector<ImageRequest> request{ ImageRequest("0", ImageType::Scene), ImageRequest("0", ImageType::DepthPlanar, true), ImageRequest("0", ImageType::Segmentation, false, false) };
// const std::vector<ImageRequest> request{ ImageRequest("0", ImageType::Scene), ImageRequest("0", ImageType::DepthPlanar, true), ImageRequest("0", ImageType::Segmentation) };
/* const std::vector<ImageRequest> request{ ImageRequest("front_center", ImageType::Scene), ImageRequest("front_center", ImageType::DepthPlanar, true), ImageRequest("front_center", ImageType::Segmentation) };
const std::vector<ImageResponse>& response = client.simGetImages(request, drone_name);
std::cout << "# of images received: " << response.size() << std::endl;
// if (!response.size()) {
if (response.size()) {
std::cout << "Enter path with ending separator to save images (leave empty for no save)" << std::endl;
std::string path = "/home/sgari/Documents/AirSim/onboard_images";
// std::getline(std::cin, path);
for (const ImageResponse& image_info : response) {
std::cout << "Image uint8 size: " << image_info.image_data_uint8.size() << std::endl;
std::cout << "Image float size: " << image_info.image_data_float.size() << std::endl;
if (path != "") {
std::string file_path = FileSystem::combine(path, std::to_string(image_info.time_stamp));
std::cout << "file path: " << file_path << std::endl;
if (image_info.pixels_as_float) {
Utils::writePFMfile(image_info.image_data_float.data(), image_info.width, image_info.height, file_path + ".pfm");
}
else {
std::ofstream file(file_path + ".png", std::ios::binary);
file.write(reinterpret_cast<const char*>(image_info.image_data_uint8.data()), image_info.image_data_uint8.size());
file.close();
}
}
}
}
*/
std::cout << "Press Enter to arm the drone" << std::endl;
std::cin.get();
// client.enableApiControl(true);
client.enableApiControl(true, drone_name); // can select the name of the quadrotor eg "Drone1", "Drone2" etc
client.armDisarm(true);
/* auto barometer_data = client.getBarometerData("", drone_name);
std::cout << "Barometer data \n"
<< "barometer_data.time_stamp \t" << barometer_data.time_stamp << std::endl
<< "barometer_data.altitude \t" << barometer_data.altitude << std::endl
<< "barometer_data.pressure \t" << barometer_data.pressure << std::endl
<< "barometer_data.qnh \t" << barometer_data.qnh << std::endl;
auto imu_data = client.getImuData("",drone_name);
std::cout << "IMU data \n"
<< "imu_data.time_stamp \t" << imu_data.time_stamp << std::endl
<< "imu_data.orientation \t" << imu_data.orientation << std::endl
<< "imu_data.angular_velocity \t" << imu_data.angular_velocity << std::endl
<< "imu_data.linear_acceleration \t" << imu_data.linear_acceleration << std::endl;
auto gps_data = client.getGpsData("",drone_name);
std::cout << "GPS data \n"
<< "gps_data.time_stamp \t" << gps_data.time_stamp << std::endl
<< "gps_data.gnss.time_utc \t" << gps_data.gnss.time_utc << std::endl
<< "gps_data.gnss.geo_point \t" << gps_data.gnss.geo_point << std::endl
<< "gps_data.gnss.eph \t" << gps_data.gnss.eph << std::endl
<< "gps_data.gnss.epv \t" << gps_data.gnss.epv << std::endl
<< "gps_data.gnss.velocity \t" << gps_data.gnss.velocity << std::endl
<< "gps_data.gnss.fix_type \t" << gps_data.gnss.fix_type << std::endl;
auto magnetometer_data = client.getMagnetometerData("",drone_name);
std::cout << "Magnetometer data \n"
<< "magnetometer_data.time_stamp \t" << magnetometer_data.time_stamp << std::endl
<< "magnetometer_data.magnetic_field_body \t" << magnetometer_data.magnetic_field_body << std::endl;
// << "magnetometer_data.magnetic_field_covariance" << magnetometer_data.magnetic_field_covariance // not implemented in sensor
// get the ground truth pose of quadrotor
auto quadrotor_pose = client.simGetGroundTruthKinematics(drone_name);
std::cout << "Ground truth pose data \n"
<< "quadrotor_pose.pose.position " << quadrotor_pose.pose.position << std::endl
<< "quadrotor_pose.pose.orientation " << quadrotor_pose.pose.orientation << std::endl
<< "quadrotor_pose.twist.angular " << quadrotor_pose.twist.angular << std::endl;
*/
std::cout << "Press Enter to takeoff" << std::endl;
std::cin.get();
float takeoff_timeout = 5;
client.takeoffAsync(takeoff_timeout, drone_name)->waitOnLastTask();
// switch to explicit hover mode so that this is the fall back when
// move* commands are finished.
std::this_thread::sleep_for(std::chrono::duration<double>(5));
client.hoverAsync(drone_name)->waitOnLastTask();
std::cout << "Press Enter to fly in a 10m box pattern at 3 m/s velocity" << std::endl;
std::cin.get();
// moveByVelocityZ is an offboard operation, so we need to set offboard mode.
client.enableApiControl(true,drone_name);
auto position = client.getMultirotorState(drone_name).getPosition();
float z = position.z(); // current position (NED coordinate system).
constexpr float speed = 3.0f;
constexpr float size = 10.0f;
constexpr float duration = size / speed;
DrivetrainType drivetrain = DrivetrainType::ForwardOnly;
YawMode yaw_mode(true, 0);
std::cout << "moveByVelocityZ(" << speed << ", 0, " << z << "," << duration << ")" << std::endl;
client.moveByVelocityZAsync(speed, 0, z, duration, drivetrain, yaw_mode, drone_name);
std::this_thread::sleep_for(std::chrono::duration<double>(duration));
std::cout << "moveByVelocityZ(0, " << speed << "," << z << "," << duration << ")" << std::endl;
client.moveByVelocityZAsync(0, speed, z, duration, drivetrain, yaw_mode, drone_name);
std::this_thread::sleep_for(std::chrono::duration<double>(duration));
std::cout << "moveByVelocityZ(" << -speed << ", 0, " << z << "," << duration << ")" << std::endl;
client.moveByVelocityZAsync(-speed, 0, z, duration, drivetrain, yaw_mode, drone_name);
std::this_thread::sleep_for(std::chrono::duration<double>(duration));
std::cout << "moveByVelocityZ(0, " << -speed << "," << z << "," << duration << ")" << std::endl;
client.moveByVelocityZAsync(0, -speed, z, duration, drivetrain, yaw_mode, drone_name);
std::this_thread::sleep_for(std::chrono::duration<double>(duration));
client.hoverAsync(drone_name)->waitOnLastTask();
std::cout << "Press Enter to land" << std::endl;
std::cin.get();
float land_timeout = 10;
client.landAsync(land_timeout, drone_name)->waitOnLastTask();
std::cout << "Press Enter to disarm" << std::endl;
std::cin.get();
client.armDisarm(false, drone_name);
}
catch (rpc::rpc_error& e) {
const auto msg = e.get_error().as<std::string>();
std::cout << "Exception raised by the API, something went wrong." << std::endl
<< msg << std::endl;
}
return 0;
}
Question
C++ example for Airsim multi-quadrotor scenario like in multi_agent_drone.py
What's your question?
I am trying to simulate multiple quadrotors in Airsim using C++, just like the example shown in AirSim/PythonClient/multirotor /multi_agent_drone.py. I found the AirSim/HelloDrone /main.cpp example as a starting point but I have questions on getting a working C++ multi-quadrotor simulation setup in Unity using MultiDroneDemo.unity.
Include context on what you are trying to achieve
Run multi-quadrotor simulation with Airsim in Unity with C++ Airsim library, like the example in AirSim/PythonClient/multirotor /multi_agent_drone.py; I want to get the IMU, RGB frame, semantic segmentation, and depth image data from each quadrotor.
Context details
OS: Linux Ubuntu 18.04 LTS Python version: 3.8 C++ version: 14 Unity: Unity editor 2021.3.31f1
settings.json includes multiple quadrotors according to documentation here.
Include details of what you already did to find answers
3128 Support for multiple drones in Unity
1540 Multiple drones in AirSim Unity?
839 Controlling Multiple Drones through ROS
706 Enabling multi-drone control
664 Update multi_vehicle.md with python API example