David McPherson

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2017-206

December 13, 2017

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-206.pdf

Nature’s self-organizing teams have long inspired engineering researchers. Imitating birds and fish, researchers built swarms that could keep formation via elegant decentralized control algorithms. Inspired by ants, they applied formation keeping to carrying loads as a mobile robot team. In this work, we further specialize this collaborative manipulation to manipulating teammates, thereby creating collaborative locomotion. Our design space is enriched by designing both the manipulator and the manipulatee at the same time. Traditional collaborative manipulation work focuses on manipulating objects in the same plane of movement as the mobile robots and adheres strictly to form or force closure manipulation paradigms. Our new design richness allows us to leverage work on in-hand, dexterous manipulation and dynamic manipulation techniques to manipulate objects (now teammates) in degrees of freedom (DOF) outside our plane of movement. These further manipulation DOF are accessed by especial care in designing the contact surfaces between robots which allows us to guide objects while slipping, instead of rigidly demanding complete closure. This paper will demonstrate the collaborative locomotion concept in the simplest possible cooperative maneuver: rolling a pronated robot back onto its feet (adding a previously inaccessible roll DOF to a differentially driven robot). Within this showcase application we discover several possible designs for cooperative righting strategies. The strategies follow the traditional manipulation classification framework: kinematic, quasi-static, and dynamic. We investigate each method analytically as well as implementing the best method on real robots.

Advisors: Ronald S. Fearing

---

BibTeX citation:

@mastersthesis{McPherson:EECS-2017-206,
    Author= {McPherson, David},
    Editor= {Fearing, Ronald S.},
    Title= {Collaborative, Contact-based Schemas for Robot Righting utilizing Hull Shape Design},
    School= {EECS Department, University of California, Berkeley},
    Year= {2017},
    Month= {Dec},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-206.html},
    Number= {UCB/EECS-2017-206},
    Abstract= {Nature’s self-organizing teams have long inspired engineering researchers. Imitating birds and fish, researchers built swarms that could keep formation via elegant decentralized control algorithms. Inspired by ants, they applied formation keeping to carrying loads as a mobile robot team. In this work, we further specialize this collaborative manipulation to manipulating teammates, thereby creating collaborative locomotion. Our design space is enriched by designing both the manipulator and the manipulatee at the same time. Traditional collaborative manipulation work focuses on manipulating objects in the same plane of movement as the mobile robots and adheres strictly to form or force closure manipulation paradigms. Our new design richness allows us to leverage work on in-hand, dexterous manipulation and dynamic manipulation techniques to manipulate objects (now teammates) in degrees of freedom (DOF) outside our plane of movement. These further manipulation DOF are accessed by especial care in designing the contact surfaces between robots which allows us to guide objects while slipping, instead of rigidly demanding complete closure. This paper will demonstrate the collaborative locomotion concept in the simplest possible cooperative maneuver: rolling a pronated robot back onto its feet (adding a previously inaccessible roll DOF to a differentially driven robot). Within this showcase application we discover several possible designs for cooperative righting strategies. The strategies follow the traditional manipulation classification framework: kinematic, quasi-static, and dynamic. We investigate each method analytically as well as implementing the best method on real robots.},
}

EndNote citation:

%0 Thesis
%A McPherson, David 
%E Fearing, Ronald S. 
%T Collaborative, Contact-based Schemas for Robot Righting utilizing Hull Shape Design
%I EECS Department, University of California, Berkeley
%D 2017
%8 December 13
%@ UCB/EECS-2017-206
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-206.html
%F McPherson:EECS-2017-206