This paper aims to address the problem of integrating sensor feedback in robotized interior finishing operations. Its motivation is to finally realize automatic operations necessitating no human intervention. A vision-based approach is proposed for monitoring the execution status and changing the action accordingly.
First, a robotic system is proposed which can realize two typical interior finishing operations, namely, putty applying and wall sanding. Second, a new method based on a deep neural network is proposed to process the visual information capturing the execution status of the interior finishing operations. It helps to determine essential parameters on where should be processed and how to execute the corresponding operation. With the proposed method, vision information is embedded into the execution of interior finishing in a closed loop style.
The experiments demonstrate the feasibility of the proposal and reveal problems for further improvement of the autonomous interior finishing robot.
This provides an original insight into robotized interior finishing by addressing an attempt on integrating visual feedback into the manual process.
