Development of co-design frameworks for optimal variable compliant actuation

Official URL: http://risc01.sabanciuniv.edu/record=b1620504 (Table of Contents)

Robotic systems powered by variable stiffness actuators (VSAs) provide important benefits for many applications that demand safety, performance, and energy-efficiency. Being interdisciplinary systems, the performance of such devices cannot be fully utilized unless proper co-design techniques that account for the inherent design couplings by synergetically designing mechanical and control disciplines, are used for their optimization. The first part of this thesis presents development of both a simultaneous and a nested co-design framework, which are proven methods in guaranteeing system-level optimality, for robotic systems actuated by VSAs and demonstrates their applicability and effectiveness on a case study for system-level design of a robotic prosthesis driven by a Mechanically Adjustable Compliance and Controllable Equilibrium Position Actuator (MACCEPA) undergoing periodic/real-life dynamic motions for a design objective to help reduce physical system weight. Similar to the simultaneous optimization strategy, the nested co-design framework can solve for system level optima, while unlike simultaneous optimization strategies, it does so by decoupling the static and dynamic optimization problems. This decoupling not only results in better scalability due to smaller sized subproblems, but also enables generalization of optimization techniques that can be applied by allowing utilization of domain specific optimization tools. On the other hand, a simultaneous approach - due to its less elaborate structure - is advantageous for applications where implementation simplicity and convergence speed are of paramount importance. In addition to the generation of robot designs that are suboptimal, the literature also lacks an integrated approach for comparison of robots driven by compliant actuation strategies. In the second part of the thesis, a unified design comparison framework that utilizes the simultaneous co-design strategy for system-level optimization of compliantly actuated robots has been developed. The simultaneous co-design method which is utilized throughout system designs carried out guarantees compliantly-actuated robotic systems performing at their full potential, while a design comparison approach based on homotopy and Pareto optimality concepts maintains integrity among compared system designs via generation of a continuum of robot designs actuated with varying degrees of controllable and fixed compliance. The direct consequence of the developed framework is that it allows robot designers make well-informed and unbiased decisions in selecting designs from among a variety of systems which are guaranteed to perform at their best. Applicability of the introduced framework has been validated through its implementation using the case study example for system-optimal design of the active knee prosthesis driven by the MACCEPA actuator under periodic/real-life dynamical task requirements.