A two-stage DRL method with a supervised collision prediction stage improves collision-free success rates on object-goal visual navigation tasks in simulation and real rooms.
DRQN-based 3D obstacle avoidance with a limited field of view,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Collision-Aware Object-Goal Visual Navigation via Two-Stage Deep Reinforcement Learning
A two-stage DRL method with a supervised collision prediction stage improves collision-free success rates on object-goal visual navigation tasks in simulation and real rooms.