Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Morphogenetic Assembly and Adaptive Control for Heterogeneous Modular Robots

Chongxi Meng, Da Zhao, Yifei Zhao, Minghao Zeng, Yanmin Zhou, Zhipeng Wang, Bin He

TL;DR

The paper tackles reconfigurable heterogeneous modular robots by proposing a two-tier approach: a hierarchical planner to efficiently solve morphology-dependent assembly and an adaptive, morphology-agnostic motion controller based on GPU-accelerated annealed-variance MPPI. The high-level planner uses bidirectional heuristic search with a type-penalty to guide module relocations, while the low-level planner computes optimal trajectories with A*,; motion generation uses a multi-stage variance annealing strategy to balance exploration and convergence in real time. Key findings show that type-penalization improves robustness, greedy heuristics reduce execution cost, and the annealed-variance MPPI achieves real-time 50 Hz control with superior velocity tracking over standard MPPI. The results validate a full cycle from module assembly to dynamic control and suggest practical impact for scalable, adaptive modular robotics, with future work on Sim-to-Real transfer and lifecycle management using a reusable module resource pool.

Abstract

This paper presents a closed-loop automation framework for heterogeneous modular robots, covering the full pipeline from morphological construction to adaptive control. In this framework, a mobile manipulator handles heterogeneous functional modules including structural, joint, and wheeled modules to dynamically assemble diverse robot configurations and provide them with immediate locomotion capability. To address the state-space explosion in large-scale heterogeneous reconfiguration, we propose a hierarchical planner: the high-level planner uses a bidirectional heuristic search with type-penalty terms to generate module-handling sequences, while the low level planner employs A* search to compute optimal execution trajectories. This design effectively decouples discrete configuration planning from continuous motion execution. For adaptive motion generation of unknown assembled configurations, we introduce a GPU accelerated Annealing-Variance Model Predictive Path Integral (MPPI) controller. By incorporating a multi stage variance annealing strategy to balance global exploration and local convergence, the controller enables configuration-agnostic, real-time motion control. Large scale simulations show that the type-penalty term is critical for planning robustness in heterogeneous scenarios. Moreover, the greedy heuristic produces plans with lower physical execution costs than the Hungarian heuristic. The proposed annealing-variance MPPI significantly outperforms standard MPPI in both velocity tracking accuracy and control frequency, achieving real time control at 50 Hz. The framework validates the full-cycle process, including module assembly, robot merging and splitting, and dynamic motion generation.

Morphogenetic Assembly and Adaptive Control for Heterogeneous Modular Robots

TL;DR

The paper tackles reconfigurable heterogeneous modular robots by proposing a two-tier approach: a hierarchical planner to efficiently solve morphology-dependent assembly and an adaptive, morphology-agnostic motion controller based on GPU-accelerated annealed-variance MPPI. The high-level planner uses bidirectional heuristic search with a type-penalty to guide module relocations, while the low-level planner computes optimal trajectories with A*,; motion generation uses a multi-stage variance annealing strategy to balance exploration and convergence in real time. Key findings show that type-penalization improves robustness, greedy heuristics reduce execution cost, and the annealed-variance MPPI achieves real-time 50 Hz control with superior velocity tracking over standard MPPI. The results validate a full cycle from module assembly to dynamic control and suggest practical impact for scalable, adaptive modular robotics, with future work on Sim-to-Real transfer and lifecycle management using a reusable module resource pool.

Abstract

This paper presents a closed-loop automation framework for heterogeneous modular robots, covering the full pipeline from morphological construction to adaptive control. In this framework, a mobile manipulator handles heterogeneous functional modules including structural, joint, and wheeled modules to dynamically assemble diverse robot configurations and provide them with immediate locomotion capability. To address the state-space explosion in large-scale heterogeneous reconfiguration, we propose a hierarchical planner: the high-level planner uses a bidirectional heuristic search with type-penalty terms to generate module-handling sequences, while the low level planner employs A* search to compute optimal execution trajectories. This design effectively decouples discrete configuration planning from continuous motion execution. For adaptive motion generation of unknown assembled configurations, we introduce a GPU accelerated Annealing-Variance Model Predictive Path Integral (MPPI) controller. By incorporating a multi stage variance annealing strategy to balance global exploration and local convergence, the controller enables configuration-agnostic, real-time motion control. Large scale simulations show that the type-penalty term is critical for planning robustness in heterogeneous scenarios. Moreover, the greedy heuristic produces plans with lower physical execution costs than the Hungarian heuristic. The proposed annealing-variance MPPI significantly outperforms standard MPPI in both velocity tracking accuracy and control frequency, achieving real time control at 50 Hz. The framework validates the full-cycle process, including module assembly, robot merging and splitting, and dynamic motion generation.
Paper Structure (11 sections, 9 equations, 8 figures, 2 algorithms)

This paper contains 11 sections, 9 equations, 8 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. System Description
  4. Problem Statement
  5. Hierarchical Reconfiguration Planner
  6. Dynamic Motion Generation
  7. Experiments
  8. Ablation Study on Heuristic Design
  9. Impact of Heuristics on Execution Cost
  10. Controller Performance Evaluation
  11. Conclusion and Future Work

Figures (8)

  • Figure 1: Heterogeneous Modular Robotic System. The system comprises a builder robot and a suite of standardized functional modules, including rigid units and actuated joint units. The figure demonstrates a quadrupedal configuration where the builder, integrated onto the torso, actively manipulates modules to reconfigure the body structure.
  • Figure 2: Heterogeneous Modular Robotic System.The framework enables dynamic morphological transitions between different robot configurations. (Blue Arrows) Robot Combination: An assembler robot merges a quadruped and a wheeled robot into a unified wheel-legged system by manipulating functional modules. (Red Arrows) Robot Separation: The inverse process where the composite wheel-legged robot is split back into its constituent quadruped and wheeled sub-systems. The center insets illustrate the step-by-step module manipulation sequence executed by the assembler.
  • Figure 3: The three heterogeneous module types and the deployment mechanism of the wheel module: (a) The passive base module serves as the primary structural element. (b) The joint module provides a single rotational degree of freedom (DOF), $\theta^J \in [0^\circ, 360^\circ]$. (c) The wheel module enables continuous rotation ($\theta^W$) for locomotion. (d) The wheel in its retracted configuration. (e) The screw-driven reconfiguration process. and (f) The wheel in its fully deployed configuration, ready for locomotion. All modules share a common $12\,\text{cm}$ cubic form factor.
  • Figure 4: The assembler robot. (a) Overall design, featuring a dual-foot base and a multi-DOF manipulator. (b) The degrees of freedom (DOF) of the robot. (c) Demonstration of a mobile assembly task, where the robot locomotes across an assembled structure to manipulate a module.
  • Figure 5: Success rate comparison for heuristics with and without the type-penalization term. The penalty is shown to be critical for ensuring planner robustness and scalability across our diverse benchmark suite.
  • ...and 3 more figures