Review of reconfiguration methods for modular self-reconfiguring robots

Modular self-reconfigurable robots (MSRRs) have emerged as an important solution to the long-standing contradiction between the task specificity of conventional robots and the demand for high adaptability in unstructured environments. In contrast to fixed-configuration systems, MSRRs are composed of standardized modules integrating sensing, actuation, communication, and connection functions, and can repeatedly reorganize their topology to alter both morphology and functionality. This capability gives them remarkable advantages in task versatility, fault tolerance, maintainability, and cost-effective mass production, making them highly attractive for applications such as disaster rescue, deep-space exploration, nuclear facility maintenance, and other high-risk operations. However, most previous reviews have classified modular robots mainly by geometric appearance or connection topology, such as chain-type, lattice-type, and hybrid systems. Although useful, these taxonomies do not fully reveal the intrinsic differences in how reconfiguration is physically realized, and therefore cannot clearly explain why different systems exhibit distinct kinematic constraints, control burdens, structural redundancy, and application boundaries. To address this issue, this paper re-examined modular robots from the perspective of reconfiguration mechanisms, aiming to provide a clearer framework for system comparison and for the rational selection of robot architectures in future engineering practice.

From the viewpoint of reconfiguration principles, the reviewed systems are categorized into four representative classes: non-self-reconfigurable systems, mobile self-reconfigurable systems, translation/rotation self-reconfigurable systems, and joint-motion self-reconfigurable systems. Non-self-reconfigurable robots rely on human assistance or external devices to disassemble and reassemble modules. By removing complicated autonomous docking mechanisms and redundant actuators, they usually achieve high stiffness, high load-to-weight ratio, low cost, and strong reliability, which makes them suitable for precision assembly, heavy-duty carrying, flexible human-robot interaction, and harsh environments. Mobile self-reconfigurable robots use wheels, legs, propellers, or similar locomotion devices to actively travel through the environment and dock into target structures. This route is especially effective for sparse reconfiguration over large distances and has inspired a wide range of planning methods, including optimal planning, distributed recruitment, bio-inspired coordination, and learning-based assembly strategies. Yet the mobility hardware that enables long-range relocation also introduces redundancy, higher manufacturing cost, and reduced suitability for microgravity scenarios. Translation/rotation self-reconfigurable robots, which are typically lattice-based, reconfigure through discrete sliding or pivoting motions supported by neighboring modules or the environment. Their dense packing, deterministic motion, and low wear under electromagnetic actuation make them promising for large space-structure assembly, modular satellites, and programmable matter, while their limited post-reconfiguration mobility constrains dynamic operations. Joint-motion self-reconfigurable robots integrate one or more active joints within each module and combine internal articulation with connection and disconnection operations. This enables multimodal locomotion and rich shape adaptation, allowing the same system to behave as a snake. crawler, manipulator, rolling loop, or deployable space structure. At the same time, these systems generally face the greatest challenges in kinematic modeling, coordinated control, and algorithmic complexity. Through a systematic comparison of representative prototypes, this review shows that the four categories form a mechanism-driven spectrum with clear trade-offs among global mobility, local deformation capability, structural compactness, control difficulty, energy efficiency, and environmental adaptability.

The comparison demonstrates that there is no universally optimal reconfiguration mode; instead, each route corresponds to a distinct balance between mechanical simplicity, reconfiguration freedom, and functional performance. Non-self-reconfigurable systems remain advantageous when stiffness, payload capacity, and reliability are dominant concerns. Mobile systems are preferable when large-scale relocation and distributed deployment are required. Translation/rotation systems are highly competitive for compact, stable, and repeatable assembly in structured or microgravity environments. Joint-motion systems provide the richest behavioral diversity and are therefore attractive for highly adaptive missions in complex terrain and on-orbit service. Future progress is expected to focus on several directions: reducing hardware redundancy while preserving reconfiguration capability; improving docking reliability together with power, signal, and load transmission performance; developing scalable planning and control methods for large numbers of modules; strengthening embodied perception and environment-aware coordination; and promoting tighter integration between modular robots and application-specific platforms such as large manipulators, orbital servicing systems, and intelligent infrastructure. In particular, for future space missions, low-energy and high-precision reconfiguration under microgravity, combined with autonomous topology variation and cooperative manipulation, is likely to become a key research frontier. Overall, mechanism-oriented classification not only clarifies the essential differences among existing modular robots but also provides a practical basis for configuration design, strategy selection, and application optimization in next-generation modular robotic systems.