Robot Vision




Control of robots using visual information

We are developing various robot vision systems.

Vision based robot control

An autonomous robot should recognize the real-world, automatically. A video camera is one of the intelligent sensors providing a large amount of information on external environments through acquired images. In our laboratory, we develop visually controlled autonomous robot systems that recognize the real world.

A man-machine communication system

An effective man-machine communication strategy is proposed by use of multiple robots. If a person sitting on a chair or lying in the bed wishes to take an object distant on the floor, he/she points to the object. A vision system on the ceiling watches the direction and decides the object pointed. Then a mobile robot goes to the object: It takes the object, and brings it to a manipulator placed close to him/her: The manipulator grasps the object and hands it to him/her.


Distance sensing by a robot for removing a weed

Sensing a distance

Recognition of a specified weed



Visual feedback




Hovering/Tracking by an aerial robot


Spherical blimp
Tracking a moving object

Hovering control



Multi-agent system for assisting human intention