search

tudelft / AvoidBench

AvoidBench
MIT License
58 stars 13 forks source link

Implemented in real environment #8

Open lgh5054 opened 3 months ago

lgh5054 commented 3 months ago

hello

As in the paper, we are trying to implement it using PX4 in a real environment. If it's okay, is there any way I can get help?

NPU-yuhang commented 3 months ago

Which part do you want to use in the real drone? If you want to use the metrics of avoidbench, you can just put our 'avoidbench' folder in your own PX4 ros workspace (some third party packages are needed of course). Then you have to change 'avoid_manage.cpp' according to your own flight tasks.

lgh5054 commented 3 months ago

@NPU-yuhang First of all, thank you for your reply.

I am also a master's student, and I would like to transfer the avoidance path used in agile_autonomy to PX4 as in the thesis. And I would like to compare the success rate of AvoidBench simulation and real world environment. I am curious about how you controlled the PX4 aircraft using the data output from agile_autonomy.

NPU-yuhang commented 3 months ago

If you are using PX4, I would recommend the state machine from ZJU Fast-Lab Fast-Drone-250. Even though they didn't implement body rates control command, you can still use the output from Agile_autonomy (body rates) to this state machine and then send to the flight controller. Of course, you can also use the original PX4 architecture based on MAVROS.

lgh5054 commented 2 months ago

@NPU-yuhang Thank you for always. Are you sure you are talking about Fast-Drone-250 MPC? Sorry to ask you this, but if possible, can I see an example launch file?