Design and implementation of HexBot: A modular self-reconfigurable robotic system

Abstract

This paper proposes the design and implementation of a planar hexagonal modular self-reconfigurable robotic system (MSRRS). A universal module is carefully designed to reflect the primary required criteria such as homogeneity, cost-effectiveness, fast actuation, and quick strong connections. While our working prototype is both large and restricted to a planar motion, it is designed so that the hardware and software can be scaled up in the number of units and down in unit size and it can be extended to accommodate 3D applications. One of the novel approaches in our design is that it is based on a multilayer approach where each layer is dedicated to perform a specific task; in other words, the design itself is considered to be modular. This multilayer approach in both the hardware and the software enables each layer to be modified and enhanced while keeping the remaining layers untouched, providing openness, flexibility, and ease of modification. The software infrastructure of this system is designed to allow the implementation of different hierarchies for distributed control and communication.

Introduction

A conventional robot with a fixed architecture is usually designed for limited tasks and can work in a particular environment, such as [1], [2], [3]. Therefore, a given unit of this kind will no longer perform well if its tasks or its environment are changed. On the contrary, MSRRSs gained popularity due to their adaptability, as their functionalities were no longer restricted to a specific task or a certain environment. As explained in [4], many successful conventional robots are designed by mimicking the dynamical function of living creatures; however, MSRRSs are designed by mimicking the structural formation of living creatures, where every organ is composed of its fundamental components, such as cells. Although each fundamental component is quite simple in its shape, intelligence, etc., a huge combination of them can form a powerful and complex system. MSRRSs implement the same concept by being composed of several simple robotic modules, where each module has the ability to move around its neighboring modules and change its location using its primitive actuators, sensors, processors, and communication units. Therefore, the complete system would be capable of autonomously transforming into other shapes by changing the position and orientation of any of its modules.

Key features of MSRRSs, such as versatility, robustness, self-reproduction, scalability, and cost-effectiveness [5], [6], [7], [8], [9], [10], [11], [12], [13], [14] enable them to perform in a wide range of applications. For example, MSRRSs can form different locomotion gaits to move in unstructured, uncertain or dynamically changing environments and carry out search and rescue missions. They can also be used for sea or space exploration, surveillance, or operations in hazardous or remote environments. There is also an interest in using MSRRSs to form growing structures, such as 3D simulated physical parts or emergency [15].

MSRRS platforms are generally classified based on their architectural topologies including mobile, lattice, chain, and hybrid. Modules in mobile architectures are designed to be able to move independently from each other [7]. In lattice architecture, modules are more similar to biological cells and can fill discrete positions in a grid structure. In general, this configuration performs very well for reconfiguration but not for locomotion [16]. On the other hand, modules in chain architecture are connected to each other in a serial manner, forming tree (open) or loop (closed) structures, which enable them better motion generation performance for locomotion at a cost of lower reconfiguration ability [17]. Finally, hybrid architectures combine both lattice and chain configurations to take advantage of their reflective benefits while eliminating their reflective drawbacks [18]. Therefore, hybrid systems are capable of performing remarkable motion generation required for locomotion, and are also capable to reconfigure into different configurations [4], [19]. The HexBot platform described in this work has a lattice homogeneous structure, as shown in Fig. 1. Similar works in this category include the following: ATRON [19], Crystalline [20], Metamorphic [21], Molecule [22], and Telecube [23]. Furthermore, the HexBot platform is able to rearrange its shape using a reconfiguration path planner and a control algorithm, which were developed to determine the required sequence of individual module movements that transforms the shape of the system from an arbitrary initial configuration to a desired goal configuration [24]. The core of this algorithm relies on a heuristic function and a Markov Decision Process (MDP) optimization [25], [26] to perform the reconfiguration in an optimal manner while enforcing several constraints and taking into account the kinematic model of the system.