Collaborative Air-Ground Target Searching in Complex Environments
- Our conference paper is accepted by SSRR 2017
-
First follow the instructions here
-
Then install several 3rd-party ROS Packages:
sudo apt-get install ros-indigo-cv-bridge
sudo apt-get install ros-indigo-aruco
sudo apt-get install ros-indigo-camera-info-manager
sudo apt-get install ros-indigo-v4l-utils
-
Download the source file from here
-
Installation:
sudo dpkg -i cuda-repo-l4t-r21.3-6-5-prod_6.5-42_armhf.deb
sudo apt-get update
sudo apt-get install cuda-toolkit-6-5
- Set GPU to be accessible by current user:
sudo usermod -a -G video $USER
- Set the environment variable:
gedit ~/.bashrc
- Then add the following lines to ~/.bashrc:
export PATH=/usr/local/cuda-6.5/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-6.5/lib:$LD_LIBRARY_PATH
- Then source the .bashrc file again:
source ~/.bashrc
-
Download the source file from here
-
Install the dependencies:
sudo dpkg -i libopencv4tegra-repo_l4t-r21_2.4.10.1_armhf.deb
sudo apt-get update
sudo apt-get install libopencv4tegra libopencv4tegra-dev libopencv4tegra-python
sudo apt-get install libgtk2.0-dev pkg-config
-
Download the source file from here
-
Unzip:
unzip opencv-2.4.10.zip
- Compile OpenCV:
Under the parent directory of "opencv-2.4.10", make another directory named "build":
mkdir build
cd build
cmake -DWITH_CUDA=ON -DCUDA_ARCH_BIN="3.2" -DCUDA_ARCH_PTX="" -DBUILD_TESTS=OFF -DBUILD_PERF_TESTS=OFF ../opencv-2.4.10/
Take a look at the message after cmake command is completed. If Use CUDA is Yes, then CUDA can be used.
PS: If "cmake" cannot be find after typing the above commands, please install the essential applications first:
sudo apt-get install build-essential make cmake g++
- Install OpenCV:
sudo make -j4 install
- Modify the environment variables:
echo "# Use OpenCV and other custom-build libraries" >> ~/.bashrc
echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib/" >> ~/.bashrc
source ~/.bashrc
- Basic framework based on ROS (almost done)
- UGV control
- UGV onboard sensor fusion (imu+gps+encoder)
- AprilTag / Aruco Code detection & positioning (UAV / UGV 互相观测校准)
- Auto-landing?
- To be continued...
从读代码开始：
我们拿DJI Onboard SDK -- ROS Example为例，其他的例程（Qt, STM32, Command Line）大同小异。
这里面有7个repository / ROS Package，暂时先忽略掉其中5个，一开始我们从_dji_sdk_和_dji_sdk_lib_这两个入手
我们先来看_dji_sdk_lib_这个repository 顾名思义，这是DJI Onboard SDK的库（libraries），阅读这里的代码可以知道目前的SDK都提供了哪些库函数，具体可以实现什么样的功能
首先，我们点进_include_文件夹来看看头文件们
打开_DJI_API.h_, 可以看到这里定义了一个名叫_CoreAPI_的class，这是一个核心的类
接下来找到这个class里面的一个函数：getBroadcastData()；返回值类型为_BroadcastData_，这是一个包含了几乎所有飞控回发的飞行数据的结构体。那么，这个_BroadcastData_里面究竟包含了哪些数据呢？
我们打开_DJI_Type.h_来看看这个结构体的定义:
typedef struct BroadcastData {
unsigned short dataFlag;
TimeStampData timeStamp;
QuaternionData q;
CommanData a;
VelocityData v;
CommonData w;
PositionData w;
MagnetData mag;
GPSData gps; // For A3 Flight Controller only
RTKData rtk; // For A3 Flight Controller only
RadioData rc;
GimbalData gimbal;
FlightStatus status;
BatteryData battery;
CtrlInfoData ctrlInfo;
uint8_t activation;
} BroadcastData;
可以清楚地看到，里面有时间戳、四元数、位置数据、磁力计、遥控器通道、云台数据、电池电量等等信息，每类数据的具体定义也在此头文件内。
比如遥控器通道数据_RadioData_：
typedef struct RadioData
{
int16_t roll;
int16_t pitch;
int16_t yaw;
int16_t throttle;
int16_t mode;
int16_t gear;
} RadioData;
roll, pitch, yaw：横滚、俯仰、偏航杆量
throttle：油门杆量
mode：遥控器左上方的飞行模式拨杆：P / A / F
gear：遥控器右下角自动返航按键上的拨杆（原本用于控制起落架）
接下来举个栗子，如果想要获得目前遥控器通道的油门杆量，只要在自己的程序里调用这个函数：
int16_t my_throttle = coreAPI->getBroadcastData().RadioData.throttle;
（其中_coreAPI_是一个事先创建的pointer,类型为_CoreAPI*_）
常见Bug汇总：
-
收不到M100飞控回发数据：
--- PC调参软件DJI Assistant 2里 “启用API控制”未勾选 -
无法使用SDK控制飞行器：
--- 多种可能性：- PC调参软件DJI Assistant 2里 “启用API控制”未勾选
- 遥控器档位未拨到F档
- 未夺取控制权（Obtain Control）
-
DJI Go App不提示激活（新设备）：
--- 换一个移动设备（iOS换成Android, vice versa...） -
Guidance不工作：
--- Guidance固件并非最新：http://www.dji.com/cn/product/guidance/info#downloads
（个人感觉90%来自新用户报告的Guidance无法使用的bug都是因为固件没升级……首先确保PC端Guidance调参软件的版本为最新，再进入调参软件里检查Guidance固件是否为最新，因为Guidance调参软件是不能自己更新的…必须从官网下载最新的安装）