Closing the Loop Using Visual Sensors

Controlled active vision is the use of a vision sensor for the provision of state information that is used in the feedback loop. Unlike traditional sensors such as rate gyros, compasses, accelerometers, etc. a visual sensor requires additional processing of the sensor data to actually provide the necessary state information for feedback. The algorithm used to extract the state information can play an important role with regards to stability of the closed-loop system or can be affected by the control actuation computed from the state feedback. The research here studies the role of the visual sensor in the feedback loop and identifies ways to improve the stability of the closed-loop system or to ensure robustness of the visual processing algorithm to any disturbances that are induced by the feedback control law or to performance variation induced by the imaging process.

Members Involved: Peter Karasev, Miguel Serrano.

Support: This work was supported in whole or in part by and NSF (0846750).