automate's profile - activity

Hi, I am trying to implement autonomous navigation with obstacle avoidance for quadcopter (which has Pixhawk). All I want to implement is quadcopter being able to move from point A to point B while avoiding obstacles. I am fine with the navigation being in 2D.

I am using Odroid connected to Pixhawk as the main controller. I am using Hokuyo URG-04LX-UG01 Scanning Laser Rangefinder for detecting obstacles.

What I figured out till now is there are two options to achieve navigation with obstacle avoidance. Use MAVROS somehow along with Navigation Stack / MoveIt. I guess as I am fine with navigation in 2D Navigation Stack should be what I should use. Is that correct?

Navigation stack requires 3 sort of data to function: Sensor Data, TF & Odometry data. I have LIDAR publishing sensor data, so thats fine.

My questions are:

1. How do I get TF and Odometry data (used by navigation stack) using MAVROS? What sort of data should I use? How do I do that?
2. Can I use MoveIt if I dont have map or anything and only 2D LIDAR data?
3. In future I want UAV to navigate in 3D. If answer to 2nd question is yes then should I switch to MoveIt instead? And any links for resources to read specifically about moveit and pixhawk will be very helpful.

Thanks!

I use Optitrack motion capture camera and push position data from Motive to LPE estimator to PX4. I get mocap timeout, xy timeout, sonar timeout msg and the software re initializes the position of the uav. Any idea what is the issue?

I use mavros to connect to UAV - roslaunch mavros px4.launch fcu_url:=/dev/ttyUSB0:921600

I use mocap_optitrack to get data from Optitrack Motive to ROS. However the topic published using mocap_optitrack has no timestamp. px4 locap position estimator takes difference between old time stamp and new one to see if mocap data is valid. However as I have no timestamp on /mavros/mocap/pose topic I get mocap timeout msgs and locap position reinitialized. How do I add timestamps in topic published by mocap_optitrack

I get the following error while trying to run PX4 Firmware Simulation using make posix_sitl_default gazebo

make[5]: warning: jobserver unavailable: using -j1. Add `+' to parent make rule. [ 2%] Running C++ protocol buffer compiler on msgs/SensorImu.proto [ 4%] Running C++ protocol buffer compiler on msgs/CommandAttitudeThrust.proto [ 6%] Built target sdf quaternion.proto[ 8%] Running C++ protocol buffer compiler on msgs/CommandRateThrust.proto : File not found. vector3d.proto: File not found. SensorImu.proto: Import "quaternion.proto" was not found or had errors. SensorImu.proto: Import "vector3d.proto" was not found or had errors. SensorImu.proto[ 10%] Running C++ protocol buffer compiler on msgs/CommandMotorSpeed.proto :8:12: "gazebo.msgs.Quaternion" is not defined. SensorImu.proto:10:12: "gazebo.msgs.Vector3d" is not defined. SensorImu.proto:12:12: "gazebo.msgs.Vector3d" is not defined. make[8]: * [SensorImu.pb.cc] Error 1 make[8]: Waiting for unfinished jobs.... [ 12%] Running C++ protocol buffer compiler on msgs/CommandRollPitchYawrateThrust.proto make[7]: [CMakeFiles/mav_msgs.dir/all] Error 2 make[6]: [all] Error 2 make[5]: [gazebo_build] Error 2 make[4]: [src/firmware/posix/CMakeFiles/gazebo] Error 2 make[3]: [src/firmware/posix/CMakeFiles/gazebo.dir/all] Error 2 make[2]: [src/firmware/posix/CMakeFiles/gazebo.dir/rule] Error 2 make[1]: [gazebo] Error 2 make: * [posix_sitl_default] Error 2

It looks like it cannot find quaternion.proto and vector2d.proto. how can i get those files?

I wish to send motion capture pose data of uav to px4 autopilot local position estimator. I am trying to use mocap_pose_estimator mavros pluigin. However I see https://github.com/mavlink/mavros/tre... its mentioned to use vision_pose_estimate. So which one do I need to use?

I publish mocap data from Optitrack software to mavros/mocap/pose and tf data on mavros/mocap/tf as per http://wiki.ros.org/mavros_extras documentation. Running px4.launch I see FCU: [inav] Mocap data valid msg. However I mavros/global_position/global topic does not publish anything. The local_position topic publishes nothing. As a result of that I cannot switch to auto_takeoff mode which I guess requires some source publishing global position data.

Any idea what I am missing?

I though global position data will only be published after I arm the uav. So I armed in altitude control mode. but it doesnt work. Do I need to be in off_board mode always?

I have tried publishing mocap data to /mocap/pose, /mocap/tf, mavros/vision_pose/pose but it doesnt work.

I also get no gps_fix warning. Is that an issue? If yes then how do i disable it.

I didnt find a way to convert /cmd_vel output to position command. I dotn think thats possible too. I used position control where ever possible. Your idea of using trajectory path to generate way points and feed them into position control sounds like a good idea

I edited apm_pluiginlists.yaml file. I moved mocap_pose_estimate from blacklist list to whitelist. As a result of that now on running mavros I can see "Pluigin mocap_pose_estimate" loaded and initialized. However I dont yet get it. What is the next step? Plugin needs to subscribe to pose and tf.How

Thanks for the reply @tonybaltovski. But I dont get it completely. I added mocap_pose_estimate in the px4flow_pluiglists.yaml file and launched the px4flow.launch. I do not think that is the right way. Can you please explain in detail on how to use the mocap_pose_estimate pluigin?

I wish to send Optitrack mocap pose data (using vrpn_client_ros) to Pixhawk FCU through mavros_extras - mocap_pose_estimate plugin. However I do not see an executable made for mocap_pose_estimate plugin. So how do I use it ?

How to use that mocap_pose_estimate? Do I need to add executables for that in the CMakeLists.txt?

Hi, I am using navigation stack for 2d navigation of UAV. I am using Hokuyo Lidar for obstacle detection. When the laser scan sees an obstacle it marks the points in costmap as obstacle. However if during the later scans if it doesnot find obstacle in the same spot it doesnt clear the obstacles from the costmap.

I havemarking: true, clearing: true, obstacle_range:2 and raytrace_range:20 in my costmap_common_params.yaml file. By these parameters I assume when no obstacles are seen by laserscan it shold clear area of 20m in front of it. But that doesnt happen. It clears around 1.5m area in front of it no matter what the raytrace_range is. Any insight as to what am I missing or what can i do will be very helpful.

I am using ROS Indigo.

Thanks. Please find below my costmap param files

costmap_common_params.yaml

obstacle_range: 2
raytrace_range: 20
footprint: [[-0.2,-0.2],[-0.2,0.2], [0.2, 0.2], [0.2,-0.2]]
#robot_radius: ir_of_robot
inflation_radius: 0.2

observation_sources: scan

scan: {sensor_frame: /laser, data_type: LaserScan, topic: /scan_filtered, marking: true, clearing: true}

global_costmap_params.yaml

global_costmap:
  global_frame: /map
  robot_base_frame: /base_link
  update_frequency: 1.0
  publish_frequency: 0.5
  static_map: false
  transform_tolerance: 5.0
  static_map: false
  rolling_window: true
  width: 50.0
  height: 50.0

local_costmap_params.yaml

local_costmap:
  global_frame: /map
  robot_base_frame: /base_link
  update_frequency: 50.0
  publish_frequency: 50.0
  static_map: false
  rolling_window: true
  width: 8.0
  height: 8.0
  resolution: 0.1
  transform_tolerance: 5.0
  planner_frequency: 5.0
  planner_patience: 5.0

@vooon can u please expand on your suggestion of using dronekit with mavros? Dronekit will connect to the solo and control it. What do you mean by tieing telemetry channel to network?