Kasper Stoy

Self-Reconfiguration Using Directed Growth

Self-reconfigurable robots are built from modules which are autonomously able to change the way they are connected, thus changing the overall shape of the robot. This process is difficult to control because it involves the distributed coordination of large numbers of identical modules connected in time-varying ways.

Using role-based control to produce locomotion in chain-type self-reconfigurable robots

This paper presents a role-based approach to the problem of controlling locomotion of chain-type self-reconfigurable robots. In role-based control, all modules are controlled by identical controllers. Each controller consists of several playable roles and a role-selection mechanism. A role represents the motion of a module and how it synchronizes with connected modules. A controller selects which role to play depending on the local configuration of the module and the roles being played by connected modules. We use role-based control to implement a sidewinder and a caterpillar gait in the CONRO self-reconfigurable robot. The robot is made from up to nine modules connected in a chain. We show that the locomotion speed of the caterpillar gait is constant even with loss of 75% of the communication signals. Furthermore, we show that the speed of the caterpillar gait decreases gracefully with a decreased number of modules. We also implement a quadruped gait and show that without changing the controller the robot can be extended with an extra pair of legs and produce a hexapod gait. Based on these experiments, we conclude that role-based control is robust to signal loss, scales with an increased number of modules, and is a simple approach to the control of locomotion of chain-type self-reconfigurable robots.

Using cellular automata and gradients to control self-reconfiguration

Abstract Self-reconfigurable robots are built from modules, which are autonomously able to change the way they are connected. Such a robot can, through this self-reconfiguration process, change its shape. The process has proved to be difficult to control, because it involves control of a distributed system of mechanically coupled modules connected in time-varying ways. In this paper we present an approach to the control problem where the desired configuration is grown from an initial seed module. Seeds produce growth by creating a gradient in the system, using local communication, which spare modules descend to locate the seed. The growth is guided by a cellular automaton, which is automatically generated on the basis of a three-dimensional CAD model or a mathematical description of the desired configuration. The approach is evaluated in simulation and we find that the self-reconfiguration process always converges and the time to complete a configuration scales approximately linearly with the number of modules. However, an open question is how the simulation results transfer to a physically realized self-reconfigurable robot.

Mechanical design of odin, an extendable heterogeneous deformable modular robot

Highly sophisticated animals consist of a set of heterogenous modules decided by nature so that they can survive in a complex environment. In this paper we present a new modular robot inspired by biology called Odin. The Odin robot is based on a deformable lattice and consists of an extendable set of heterogeneous modules. We present the design and implementation of a cubic closed-packed (CCP) joint module, a telescoping link, and a flexible connection mechanism. The developed robot is highly versatile and opens up for a wide range of new research in modular robotics.

Self-repair through scale independent self-reconfiguration

Self-reconfigurable robots are built from modules, which are autonomously able to change the way they are connected, thus changing the overall shape of the robot. This self-reconfiguration process is difficult to control, because it involves the distributed coordination of large numbers of identical modules connected in time-varying ways. We present an approach where a desired shape is grown based on a scalable representation of the desired configuration, which is automatically generated from a 3D CAD model. The size of the configuration is adjusted continually to match the number of modules in the system. This has the advantage that if modules are removed or added, the system automatically adjusts its scale and thus self-repair is obtained as a side effect. This capability is achieved by distributed, local rules for module movement that are independent of the goal configuration. We compare the scale independent approach to one where the desired configuration is grown directly at a fixed scale. We find that the features of the scale independent approach come at the expense of an increased number of moves, messages, and time steps taken to reconfigure.

A unified simulator for Self-Reconfigurable Robots

Generic simulation platforms such as player/stage are an essential tool in mobile robotics, but until now no similar platforms have been available for the field of self-reconfigurable robots. We here present a generic simulation platform for modular, self-reconfigurable robots: the unified simulator for self-reconfigurable robots (USSR). USSR is based on a physics engine, allowing simulation of both self-reconfiguration and dynamic interaction with the environment. The simulator is implemented as a framework that provides numerous components that can be combined to form new or existing modular robots, allowing easy experimentation: USSR currently includes support for the ATRON, Odin, and M-TRAN modular robots.

Selecting a meta-module to shape-change the ATRON self-reconfigurable robot

The ATRON self-reconfigurable robot consists of simple one degree of freedom ATRON modules. The motion capabilities of an individual module are therefore quite limited. To compensate for this, meta-modules composed of more than one module are used to shape-change the system. Meta-modules emerge from the environment created by other modules, move on the surface of other modules and stop at a new position. The flow of meta-modules, from one place to another on the structure of modules, realizes the desired self-reconfiguration. In this paper we compare six different meta-module types composed of ATRON modules. Variations of meta-module morphology and meta-actions are investigated for its ability to shape-change the robot. We conclude that two of the investigated meta-module types are able to shape-change the robot to an acceptable extent

A distributed and morphology-independent strategy for adaptive locomotion in self-reconfigurable modular robots

In this paper, we present a distributed reinforcement learning strategy for morphology-independent life-long gait learning for modular robots. All modules run identical controllers that locally and independently optimize their action selection based on the robots velocity as a global, shared reward signal. We evaluate the strategy experimentally mainly on simulated, but also on physical, modular robots. We find that the strategy: (i) for six of seven configurations (3-12 modules) converge in 96% of the trials to the best known action-based gaits within 15 min, on average, (ii) can be transferred to physical robots with a comparable performance, (iii) can be applied to learn simple gait control tables for both M-TRAN and ATRON robots, (iv) enables an 8-module robot to adapt to faults and changes in its morphology, and (v) can learn gaits for up to 60 module robots but a divergence effect becomes substantial from 20-30 modules. These experiments demonstrate the advantages of a distributed learning strategy for modular robots, such as simplicity in implementation, low resource requirements, morphology independence, reconfigurability, and fault tolerance.

A simple approach to the control of locomotion in self-reconfigurable robots

In this paper we present role-based control which is a general bottom-up approach to the control of locomotion in self-reconfigurable robots. We use role-based control to implement a caterpillar, a sidewinder, and a rolling track gait in the CONRO self-reconfigurable robot consisting of eight modules. Based on our experiments and discussion we conclude that control systems based on role-based control are minimal, robust to communication errors, and robust to reconfiguration.

Anatomy-based organization of morphology and control in self-reconfigurable modular robots

In this paper, we address the challenge of realizing full-body behaviors in scalable modular robots. We present an experimental study of a biologically inspired approach to organize the morphology and control of modular robots. The approach introduces a nested hierarchy that decomposes the complexity of assembling and commanding a functional robot made of numerous simple modules. The purpose is to support versatility, scalability, and provide design abstraction. The robots we describe incorporate anatomy-inspired parts such as muscles, bones, and joints, and these parts in turn are assembled from modules. Each of those parts encapsulates one or more functions, e.g., a muscle can contract. Control of the robot can then be cast as a problem of controlling its anatomical parts rather than each discrete module. To validate this approach, we perform experiments with micron-scale spherical catom modules in simulation. The robots we simulate are increasingly complex and include snake, crawler, quadruped, cilia surface, arm-joint-muscle, and grasping robots. We conclude that this is a promising approach for future microscopic many-modules systems, but also that it is not applicable to relatively weak and slow homogeneous systems such as the centimeter-scale ATRON.