Demo Task with Trained Model

How to run demo task with trained model

Open demo_task.yaml and change the parameters in wandb field.

...
wandb:
  entity: finani
  project: omnidrones-public
  artifact_name: Hover-ppo
  artifact_version: rtx4090-20m-n128
...

Single-Agent Tasks

OmniDrones Version	Task	Algorithm	GPU	Total Frames	# Envs	Eval Interval	Save Interval	Runtime	-	Entity	Project	Artifact Name	Artifact Version	Wandb Link
0.1.0	Hover	PPO	RTX 4090	20M	128	1000	2000	00h 55m 46s	-	finani	omnidrones-public	Hover-ppo	rtx4090-20m-n128	gptmuz5h
0.1.0	Track	PPO	RTX 4090	20M	128	1000	2000	00h 51m 37s	-	finani	omnidrones-public	Track-ppo	rtx4090-20m-n128	5ma07rww
0.1.0	FlyThrough	PPO	RTX 4090	5M	32	1000	2000	00h 58m 02s	-	finani	omnidrones-public	FlyThrough-ppo	rtx4090-5m-n32	zksd1tis
0.1.0	PayloadHover	PPO	RTX 4090	20M	128	1000	2000	00h 55m 05s	-	finani	omnidrones-public	PayloadHover-ppo	rtx4090-20m-n128	some26mz
0.1.0	PayloadTrack	PPO	RTX 4090	20M	128	1000	2000	00h 54m 39s	-	finani	omnidrones-public	PayloadTrack-ppo	rtx4090-20m-n128	p3wxnx1e
0.1.0	PayloadFlyThrough	PPO	RTX 4090	5M	32	1000	2000	00h 55m 24s	-	finani	omnidrones-public	PayloadFlyThrough-ppo	rtx4090-5m-n32	amd3nx6p
0.1.0	InvPendulumHover	PPO	RTX 4090	20M	128	1000	2000	00h 57m 01s	-	finani	omnidrones-public	InvPendulumHover-ppo	rtx4090-20m-n128	rpctr6t1
0.1.0	InvPendulumTrack	PPO	RTX 4090	20M	128	1000	2000	00h 59m 06s	-	finani	omnidrones-public	InvPendulumTrack-ppo	rtx4090-20m-n128	xy17mm2i
0.1.0	InvPendulumFlyThrough	PPO	RTX 4090	5M	32	1000	2000	00h 58m 17s	-	finani	omnidrones-public	InvPendulumFlyThrough-ppo	rtx4090-5m-n32	i193m1z3
0.1.0	Forest	PPO	RTX 4090	20M	128	1000	2000	01h 01m 10s	-	finani	omnidrones-public	Forest-ppo	rtx4090-20m-n128	pq30coja
0.1.0	Pinball	PPO	RTX 4090	5M	32	1000	2000	00h 54m 44s	-	finani	omnidrones-public	Pinball-ppo	rtx4090-5m-n32	xdjbajka

Multi-Agent Tasks

OmniDrones Version	Task	Algorithm	GPU	Total Frames	# Envs	Eval Interval	Save Interval	Runtime	-	Entity	Project	Artifact Name	Artifact Version	Wandb Link
0.1.0	PlatformHover	PPO	RTX 4090	300K	2	1000	2000	00h 54m 27s	-	finani	omnidrones-public	PlatformHover-ppo	rtx4090-300k-n2	xvp9jgzn
0.1.0	PlatformTrack	PPO	RTX 4090	300K	2	1000	2000	00h 55m 29s	-	finani	omnidrones-public	PlatformTrack-ppo	rtx4090-300k-n2	sdcl0vhm
0.1.0	PlatformFlyThrough	PPO	RTX 4090	300K	2	1000	2000	00h 54m 34s	-	finani	omnidrones-public	PlatformFlyThrough-ppo	rtx4090-300k-n2	7w2ovagr
0.1.0	TransportHover	PPO	RTX 4090	150K	1	1000	2000	00h 59m 23s	-	finani	omnidrones-public	TransportHover-ppo	rtx4090-150k-n1	6uv6ao6s
0.1.0	TransportTrack	PPO	RTX 4090	150K	1	1000	2000	00h 55m 43s	-	finani	omnidrones-public	TransportTrack-ppo	rtx4090-150k-n1	7keik7j3
0.1.0	TransportFlyThrough	PPO	RTX 4090	150K	1	1000	2000	00h 55m 08s	-	finani	omnidrones-public	TransportFlyThrough-ppo	rtx4090-150k-n1	7hs3ujd7

How to add trained model (for contributors)

• Trained model is required

  • to be uploaded to wandb public project because the model can be downloaded.

  • to be including videos (eval_interval != -1)