On radiation-based thermal servoing : perception, modelling, control, and experiments

Abstract

Robotic thermal servoing (TS) is a new sensor-based temperature control method that regulates heat energy transfer processes by actively changing the robot configuration. This control method is essential for creating machines with thermo-motor intelligence for industrial, surgical, exploration, and rescue applications. Despite its practical benefits, state-of-the-art methods typically address this problem in an openloop fashion (i.e. with no thermal feedback) and with static source-surface configurations (i.e. no robot controls). The main challenge is to derive a geometrical-thermal-motor model that describes the relation between the active robot configuration and the produced dynamic thermal response. The general objective of our research is to implement different types of TS techniques and explore their practical applications. Specifically, this thesis focuses on the robotic TS scenarios where the heat radiation is dominant. We started by investigating devices that enable the robot to perceive temperature of its surrounding environment. To this end, an RGB-depth- thermal camera system that registers an object's temperature profile to its geometrical information was developed to obtain abundant multimodal feedback in real-time. Then, we elucidates the formulation of a "fire-to-hand" robotic TS problem, where multiple objects (with unknown thermophysical properties) attached to the same robot end-effector are controlled to move around a radiative heat source. This experimental setup is a generalization of many practical industrial applications. To effectively and simultaneously regulate the temperature values the objects, two asymptotically stable controllers, one model-based and one adaptive, were designed. The experimental results validate the feasibility of our proposed method, and the unfeasible temperature target problem is analysed in depth. Next, we explored a "fire-in-hand" robotic TS problem, where a robot system that autonomously tracks the sun and concentrates the solar power through a Fresnel lens was developed. Relying on the optical simulations and the heat pyrolysis model, we could actively regulate the temperature of the target surface and the induced alteration of the surface property by robot motion. This functional robotic platform is a prototype of a new type of light-weight and energy-efficient field robot that effectively utilizes solar power for electricity generation and high temperature operations.