Robotics and Automation Expert
download pdf

An Experimental Investigation of
Robot Actuator Performance

Robert 0. Ambrose, Delbert Tesar, Richard N. Hooper
University of Texas at Austin, Mechanical Engineering, Austin, Texas, USA 78712

Forward added by Rich in 2022
Rob Ambrose wrote this paper (with a little help from me) thirty years ago and it is still just as relevant today. If you are even thinking of designing a serial robot, read this paper and understand the compliance at the robot's joints will dominate the performance of the robot. It's amazing how many adults teaching kids about designing robots don't understand this.

Abstract
Robot designers face the paradox of strength and response when selecting actuators for their
systems. Within the actuators, motors, reducers, sensors and brakes all add inertia, compliance and
friction to the drive trains. While some combinations of these components have been found to offer
superior strength-to-weight ratios, they typically achieve this performance at the expense of dynamic
response. The most demanding applications will be those that simultaneously require strength (high
payloads), response (bandwidth) and a lightweight design. While over 40 criteria have been
identified as significant in the selection of actuators [2], those that pertain specifically to the
system's dynamic response are compliance, resistance, friction, inductance and inertia. These are
recognized as the foundations of any electromechanical system, and are used as the basis for an
experimental investigation of actuator performance for robotic applications.
Keywords: Design, Actuators, Dynamics.

Introduction
The recent trend toward modular robot design is a response to manufacturing requirements f or versatility on the factory floor. Non-manufacturing operations in space and other hazardous environments also require a broader spectrum of robot performance than possible with a single manipulator.

Actuators are a central technology that drives robot design, and thus offer a logical selection as the modular level of aggregation in robotic systems [2]. Defining the actuator as a prime building block in a robot architecture separates the power and performance issues of the servo systems from the kinematics and workspace issues associated with the robot's structural design [ 19]. 

Much work has been done at the kinematics level in modular robot design. The RMMS at CMU [19][20] and the CEBOT work by Fukuda of Japan ([11] have forced their designers to formulate design methodologies for matching robot structures to specific tasks [16][12]. Beyond these kinematics methods, a robot's dynamic performance must also be matched to the mechanical requirements of a given task [2]. These requirements include an actuator's stiffness, friction, backlash, hysteresis, weight, torque, speed and bandwidth.


R. Ambrose, D. Tesar, R. Hooper. “An Experimental Investigation of Robot Actuator Performance.” Proceedings of the Second International Symposium on Measurement and Control in Robotics, November 1992, Pages: 623 – 630.