Realistic long-horizon tasks like image-goal navigation involve exploratory
and exploitative phases. Assigned with an image of the goal, an embodied agent
must explore to discover the goal, i.e., search efficiently using learned
priors. Once the goal is discovered, the agent must accur
我们提出了一种混合导航方法,将多对象导航(Multi-ON)任务分解为两个不同的技能:(1)使用经典 SLAM 和符号规划器处理航路点导航,而(2)使用结合监督学习和强化学习训练的深度神经网络处理探索、语义建图和目标检索,我们展示了该方法在模拟和真实环境中相对于端到端方法的优势,并超越了该任务的最先进技术。