Awakened structures

A Novel Robot-Material System for Parallel Construction

2023

Keywords

Collective Robotic Construction
Programmable Matter
Reconfigurable Assembly
Robot-Material Collaboration

Documentation

Thesis Booklet

Tools

Rhino+Grasshopper
Fusion360
ICD x ABM Framework
Arduino
3D Printing

Advisor

Samuel Leder
Lior Skoury
Philipp KragI
‍Prof. Achim Menges
Prof. Thomas Wortmann

Teammates

Chia-yen Wu

My Role

Concept Development
Material Experiment
Robotic Research
Design Computation
Prototype Production

Innovations

Material-Robot Co-Design Framework: Introduces a co-designed system where material behaviors synchronize with the design of the mobile robot, ensuring the robot to fit, transverse and manipulate the materials within the structure.

From Passive to Active Materials: Pioneers the integration of active, shape-changing materials into the material-robot assembly system, unlock new possibilities for efficient robotic assembly.

Simplified Robotic Systems: Demonstrates how embedding intelligence within materials streamlines robotic design, minimizing complexity while improving assembly efficiency.

Overview

Collective robotic construction (CRC) is an emerging approach to construction automation that employs multiple machines—often small robots with limited payloads—to assemble large-scale structures. Most research in this field has focused on systems where active robots assemble passive materials, resulting in complex, expensive, and less fault-tolerant systems. This undermines CRC’s core advantages of simplicity, affordability, and robustness.

To address this limitation, this research introduces a material-robot collaboration framework for CRC by embedding system intelligence within the materials. Specifically, the system uses shape-changing materials that transition between predefined states when actuated by robots, improving assembly efficiency while reducing robotic complexity.

Building on this concept, we developed a system that combines a Hoberman sphere-inspired linkage with a bespoke robot, demonstrating collaboration between materials and robotic assemblers. This framework consists of four integrated components:
‍
- Material-Robot Collaboration: A system that synchronizes programmable material behaviors with robotic actions based on assembly needs.
- Discrete Deployable Modules: Discrete modules designed to actively support robotic assembly through geometry transformation, alignment, and locking
- Robotic Assemblers: Custom-designed robots optimized for interacting with the shape-changing modules
- Assembly Algorithm: A coordinated algorithm for real-time sequencing and path planning, enabling autonomous and efficient construction.

Method:
Material-Robot CoDesign

Material-robot collaboration follows a coordinated design strategy that integrates the building modules (”material”), robot, and assembly process:- The building module’s shape and deployment inform the robot's functional requirements.
- The materials and robots' characteristics shape the way they collaborate, and vice versa.
- The assembly algorithm uses these characteristics to plan construction sequence and robot movements.

‍

Concept:
Material-Robot Collaboration

Assembly Requirements
‍

These requirements define the potential of construction robotics, enhancing both efficiency and flexibility in the assembly process.
‍

1. Parallel Building with multiple robots

2. Bridging structures from multiple ends

3. Cantilevering the structure from top to bottom

Locomotion Principles
‍

The assembly process relies on three essential robotic locomotion behaviors for constructing complex structures:

1. Extend or contract arms to pick up building units
2. Move building materials within the structure in 3-axis movement and rotation
3. Actuate the building unit into the designated place and lock it to the structure
‍

Development:
Deployable Building Unit

Inspired by compliant mechanisms like the Hoberman Sphere and FlexLinks, we developed a bistable modular block with two deployable joints and a material-material connector for locking.

‍

‍Modular, discrete, uniform geometry
- mass-producible, meaning affordability
- uniformly regular, enabling repeatability
‍

‍Bistable / multi-stable deployability
- predictable state configs
- self-locking
‍

Corner Joints:
‍Each joint features three pairs of opposing gears, allowing one active gear to collapse the box geometry.

Mid-Joints on Each Frame:
Equipped with springs and a pair of opposing gears, these joints activate adjacent sides when one side is actuated.

Material- Material Interface:
A self-selective connector locks units into place during deployment, with a male-to-female or female-to-male locking configuration.

The prototype of the modular deployable block includes aluminum rods, 3D-printed joints, and material-material connectors.

Development:
Robotic Assembler

The robotic arm's mechatronic design consists of three parts:

‍
1. Extension box for positioning the arm at the building block's center
2. Deployment box for bending/contracting the arm
3. Material-robot interface for attaching/detaching the arm to/from the building block during deployment.

‍

The arm, acutated by the stepper motor, extend itself to reach the center of the building unit.

The arm bends its elbow to reach the four sides of the building unit

The material-robot interface opens and closes to securely attach the robot to the four midjoints of the building unit.

Development:
Assembly Algorithm

This phase guides the robot in navigating the structure to execute tasks such as placing building units, disassembling supports, or returning to the start.

It consists of two main components:
- Pre-processing the input geometry
- Agent-Based Modeling

‍

Pro-processing Input Geometry
‍
The input mesh is voxelized to align with the agent-based modeling system.
Voxel center points are extracted to generate supports and rafts based on criteria like overhang angle and layer height.

‍

Agent-Based Modeling With A* Algorithm
‍‍
Agent-based modeling (ABM) combined with the A* algorithm enables path planning for assembly and disassembly.

ABM allows decentralized decision-making and captures emergent behaviors from multiple robots. The A* algorithm optimizes pathfinding, by efficiently identifying the shortest routes while adapting in real-time to the changing structure.

The ABM system encompasses 3 primary elements:
- Agent: Robots with distinct attributes and decision-making capabilities
- Agent's behaviors: Algorithms governing agents’ actions and interactions with the environment and other agents, including finding the shortest path for agents to reach target locations while transitioning between tasks.
- Environment: The space/context within which agents operate
‍

Target

Deployed

For-Build

Occupied

Support/Raft

Simulation: building with one robot agent

Simulation: disassembling the supports

Prototype

A 6-module prototype demonstrates a robot disassembling a module from the structure's center— an unprecedented capability in construction technology. Currently controlled by Arduino, the robot's functionality can be enhanced with sensors and an automated control system in the future.

‍

Set-up,: top view

Set-up: side view, undeployed

Set-up: side view, deployed

Entering the building module (2x faster)

Extending the arm and attaching to the module (2x faster)

Dissembling a module from the structure (2x faster)