Acumen Robot Model Series

Title Acumen Robot Model Series
Summary Build a series of increasingly sophisticated robot models in Acumen, to (1) explore mathematical formulations and (2) create tutorials and didactic examples.
Keywords Rigid-body dynamics, Acumen, Cyber-Physical System
References SCARA

Prerequisites Solid mathematical and programming skills
Supervisor Roland Philippsen, Walid Taha
Level Master
Status Internal Draft

Project Description

Robots are machines that embody sensors, actuators, and computational resources. They are thus an excellent example of cyber-physical system (CPS) and present very interesting twists for modeling and simulation tools. In particular, the equations of motion for robots that can be modeled as rigid-body trees are of a form that has been studied extensively [1].

Acumen is a domain specific language for modeling CPS. It is being developed by the Effective Modeling group to address a key challenge for accelerating innovation in this area. Simulation plays a key role in CPS design, and Acumen is a language for capturing and simulating the kind of hybrid continuous/discrete models needed to capture the behavior of cyber-physical systems.

The objective of this masters' thesis is to build up a series of increasingly sophisticated robot manipulator examples in Acumen. The aim is twofold: (1) explore which kinds of mathematical formulations need to be efficiently supported by Acumen in order to best support the robot design and prototyping process, and (2) create tutorials and didactic examples for teaching Acumen in particular, and CPS in general.

A preliminary idea for the sequence of robot examples is as follows.

  1. Point masses in a plane (serves double duty as intro to Acumen)
    • kinematic chains (pendulum, double pendulum, ...)
    • kinematic trees (stick figures)
  2. Inertias in a plane
    • kinematic chains
    • kinematic trees
    • real-world example: SCARA robot
  3. Inertias in three dimensions
    • study Denavit-Hartenberg (DH) parameters as a preparation for the following steps
    • repeat the first four examples in 3D space with DH parameters
    • model a PUMA arm, a very nice example of kinematic chain, with known kinematic and dynamic parameters
    • model a PR2 mobile manipulator, a very nice example of kinematic tree, also with known parameters


[1] R. Featherstone. Rigid Body Dynamics Algorithms. Springer, New York, 2008. ISBN 0387743146.