Home /Research /Visual homing: Surfing on the epipoles.
OTHER

Visual homing: Surfing on the epipoles.

Ronen Basri, Ehud Rivlin, Ilan Shimshoni

Year
1997
Citations
5

Abstract

We antroduce a novel method for vasual homang. Usang thas method a robot can be sent to desared po-sataons and oraentataons an 3-0 space specafied by san-gle amages taken from these posafzons. Our method determanes the path of the robot on-lane The start-zng posataon of the robot as not constrazned, and a 3-0 model of the envaronment a s not requared. The method as based on recoverang the epapolar geometry relatang fhe current amage taken by the robot and the target amage. Usang ihe epapolar geometry, most of the pn-rameters whzch specafy the dafferences in posataon and oraentataon of the camera between the two amages are recovered. However, sance not all of the parameters ran be recovered from two amages, we hace developed specafic methods to bypass these massang parameters and resolve the ambaguztaes that ezast We present two homang algorathms for two standard projectaon mod-els, weak and full perspectave. We have performed samulataons and real experaments whach demonstrate the robustness of the method and that the algorathms always converge to the target pose. 1

Keywords

Computer visionEpipolar geometryArtificial intelligenceRobustness (evolution)RobotComputer sciencePosition (finance)Image (mathematics)

Related papers

Browse all OTHER papers