Synergizing Morphological Computation and Generative Design: Automatic Synthesis of Tendon-Driven Grippers

TL;DR

A design methodology to generate linkage mechanisms for robots with morphological computation is proposed that uses a graph grammar and a heuristic search algorithm to create robot mechanism graphs that are converted into simulation models for testing the design output.

Abstract

Robots' behavior and performance are determined both by hardware and software. The design process of robotic systems is a complex journey that involves multiple phases. Throughout this process, the aim is to tackle various criteria simultaneously, even though they often contradict each other. The ultimate goal is to uncover the optimal solution that resolves these conflicting factors. Generative, computation or automatic designs are the paradigms aimed at accelerating the whole design process. Within this paper we propose a design methodology to generate linkage mechanisms for robots with morphological computation. We use a graph grammar and a heuristic search algorithm to create robot mechanism graphs that are converted into simulation models for testing the design output. To verify the design methodology we have applied it to a relatively simple quasi-static problem of object grasping. We found a way to automatically design an underactuated tendon-driven gripper that can grasp a wide range of objects. This is possible because of its structure, not because of sophisticated planning or learning.

Figures (9)

• Figure 1: The paper proposes a generative design approach that is based on interaction between a graph generated by a heuristic algorithm (shown on the left) and a simulation model based on the graph (shown in the center). To verify generated designs and justify the proposed procedure, physical prototypes were built (on the right)
• Figure 2: The vocabulary of rules: rule 1 initializes a palm P with finger dummies F, rule 2 adds a finger, replacing a finger dummy F with a group of nodes, rule 3 adds a phalanx instead of node FG, rules 4-7 terminate nodes, rule 8 removes a finger dummy F, and rule 9 stops a finger from growing
• Figure 3: Instructive example of a sequence of rules applied to generate the simplest one-fingered gripper: rule $r_1$ transforms an initial node into a palm with 6 finger dummies F; rule $r_2$ transforms one of dummy F's into a finger with a base B, joint J, and an attachment for the next link FG; rule $r_3$ adds extra phalanges, while rules $r_8$ and $r_9$ eliminate non-terminal nodes such as F and FG, finally rules $r_5$, $r_6$, and $r_7$ terminate non-terminal nodes
• Figure 4: Schematic representation of a finger with two phalanges in a horizontal state.
• Figure 5: Samples showing simulation models of generated gripper designs. Numbers next to images are values of a reward function used to evaluate the whole performance of generated graphs. Values around 10 indicate secure grasp of an object even with external force applied, values around 6 mean that an object was grasped, but was not able to handle an external force, and finally rewards around 3 means gripper have just touched an object without grasping