I have a question about the code in benchmarking_node.cpp. In the callback function "PointCloudCallback" 1000 points are sampled to query in sphere space around current pose. This is different from the paper, in which the query point is along the motion path. and now only the information of last query point is saved in "reply". Why the other query points are not saved as"reply".
I want to implement nanomap in UAV project. I have read the paper "Integrated Perception and Control at High Speed: Evaluating Collision Avoidance Maneuvers Without Maps". now I use px4 as flight control. How can I use the "reply" from nanomap to realize obstacle avoidence. Could we publish and subscribe the same msg (parameter) with mavros like in repository "quadrotor-collision-avoidance"? could the value in thes msgs also be computed by reward function in paper?
I have a question about the code in benchmarking_node.cpp. In the callback function "PointCloudCallback" 1000 points are sampled to query in sphere space around current pose. This is different from the paper, in which the query point is along the motion path. and now only the information of last query point is saved in "reply". Why the other query points are not saved as"reply".
I want to implement nanomap in UAV project. I have read the paper "Integrated Perception and Control at High Speed: Evaluating Collision Avoidance Maneuvers Without Maps". now I use px4 as flight control. How can I use the "reply" from nanomap to realize obstacle avoidence. Could we publish and subscribe the same msg (parameter) with mavros like in repository "quadrotor-collision-avoidance"? could the value in thes msgs also be computed by reward function in paper?