Abstract: Visual navigation in robotics traditionally relies on globally-consistent 3D maps or learned controllers, which can be computationally expensive and difficult to generalize across diverse ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results