What API would be available to control drone?

[frontw]

https://github.com/microsoft/AirSim-NeurIPS2019-Drone-Racing/blob/master/docs/competition_guidelines.md#how-to-participate-in-the-challenge

The simulation environment will later contain a low-level trajectory planner and feedback controller to control the dynamics of the drone to achieve the specified waypoint and velocity, thereby simplifying the interface as much as possible.

Would it be possible to control by drone velocities and attitude also?

[madratman]

the airsimneurips pip package is a mirror of airsim pip as of now, with an addition of simLoadLevel, as documented in the readme.

See the source directly to see list of supported APIs.
https://github.com/microsoft/AirSim/blob/master/PythonClient/airsim/client.py#L279

The link to airsim python API doc is already in the readme as well.

yes, moveByVelocityAsync is available, and there is moveByAngleThrottleAsync

We'll be releasing two new APIs next week:

moveOnSplineAsync(waypoint_list, bool_add_current_position_as_constraint, bool_add_current_velocity_as_constraint, max_speed, max_acceleration)

moveOnSplineVelConstraintsAsync(waypoint_list, velocity_constraints_list, bool_add_current_position_as_constraint, bool_add_current_velocity_as_constraint, max_speed, max_acceleration)

The above will serve as high level move APIs which will handle trajectory planner and trajectory tracking control for the user.

[ctoumieh]

@madratman What type of controller you will be using for the trajectory tracking?

[madratman]

It's a simple pure pursuit tracking controller. chasing both position and velocity references, and tracking errors in cross-track, along-track, z direction, and yaw (the reference yaw is obtained from velocity of the trajectory)
it spits out moveByVelocity commands, which are sent to the airsim API
I made the GIFs in the readme using moveOnSplineAsync()

Update here are the 1x and HD videos from which the GIFs were made

• Building 99
• ZhangJiaJie
• Soccer Field

Notes.verbose():
The drones are moving in their own coordinate frames defined at takeoff pose. hence there's one which is always tracking the center of the gate, and some which crash the gates now and then.
There are also slightly offset by time of start.
The previous two points create the illusion of an actual race.
The tracks in the videos are slightly different from the pak files, as they were early designs

[madratman]

@ctoumieh @frontw Just added the APIs I was talking about in the v0.1 release. Some sample code here https://github.com/microsoft/AirSim-NeurIPS2019-Drone-Racing/tree/master/baselines

Just run python generate_settings_file.py once before you try baseline_racer.py.