ayushgaud / path_planning

Quadcopter path planning using RRT* and minimum jerk trajectory generation
339 stars 99 forks source link

Aleady have a otcomap, How can I use adjust this project? #10

Open ElderWanng opened 5 years ago

ElderWanng commented 5 years ago

Hey there! Thank you for your awesome project! Recently I've built a SLAM system with stereo camera. ( I didn't use RGBD camera ) My project can out put a global octomap or points cloud incrementally, sample like this, after drone flying around the building, the VOXEL can be seen in the RVIZ.

NOW I wanna use your project to implement a avoidance function, I use VIO to estimate pose and send them to PX4 by /mavros/vision_pose/pose, and I stitch points cloud generated by stereo camera to build a global map. ( For the moment let us assume that called /map_out ) Can you show more hint? Which topics or action in path_planning I should subscribe to, publish to or modify the topic name?

ayushgaud commented 5 years ago

The reconstruction looks quite good!

Regarding the explanation please take a look at this issue . In summary, you need to create an octomap and provide it to "/octomap_binary" topic and your odometry to "/bebop2/odometry_sensor1/odometry" topic. The goal point needs to be published to /clicked_point topic while you will receive the waypoints from RRT* at topic "waypoints" and minimum jerk trajectory at "/bebop2/command/trajectory". You can use RViz to visualize everything. I also publish trajectory markers so it will look like this

yashagarwal9 commented 5 years ago

Hi Ayush, I really liked your work. I am working on building a system which can guide a visually impaired human to navigate through indoor conditions. I have generated a point cloud using ORBSlam2 and converted it into octomap. I am planning to use your implementation for the same. Can you give your inputs on it? I'm also planning to use Navigation Package of ROS but it requires 2D map which I think will work but I need to get the location of the objects like doors, stairs, etc which I may not get in 2D map. What are your suggestions?