notesum.ai

Published at November 27

Monocular Obstacle Avoidance Based on Inverse PPO for Fixed-wing UAVs

cs.RO

cs.CV

Released Date: November 27, 2024

Authors: Haochen Chai, Meimei Su, Yang Lyu, Zhunga Liu, Chunhui Zhao, Quan Pan

Arxiv: http://arxiv.org/abs/2411.18009v1

Refer to caption

Parameter	Value
Air Speed (m/s)	30
Depth Map Size ( $\mathcal{}{H},W$ )	224, 224
Reward Term Weight ( ${C_{1}},C_{2},C_{3},C_{4}$ )	30, -30, 0.5, 1.0
Flying Distance Cap $d_{max}$ (m)	1300
Learning Rate	0.0003
Gamma ( $\gamma$ )	0.95
Clip Range ( $\epsilon$ )	0.3
K Epochs	2
Batch Size	2048
Value Loss Coefficient( ${w_{1}}$ )	0.5
Entropy Loss Coefficient( ${w_{2}}$ )	0.1
Max Timesteps Per Episode	60
Max episodes	3000
State Dimension	256
Action Dimension	8