Source-linked AI summary

Mobile Robotic Fabrication at 1:1 scale: the In situ Fabricator

Markus Giftthaler, Timothy Sandy, Kathrin Dörfler, Ian Brooks, Mark Buckingham, Gonzalo Rey, Matthias Kohler, Fabio Gramazio, Jonas Buchli

arXiv:1701.03573v1cs.RO

TL;DR

The paper addresses how digitally controlled robots can perform manufacturing, assembly, and fabrication directly on construction sites despite their changing and demanding conditions. It presents the In situ Fabricator concept and IF1 prototype, integrates estimation, planning, and control, and demonstrates full-scale brick-wall construction and Mesh Mould fabrication. The work also identifies industrial-arm limitations and motivates a next-generation actuator and IF2 design.

  • Problem

    Construction sites remain difficult for autonomous robotics because they are changing, dirty, cluttered environments requiring capabilities beyond classical industrial automation.

  • Method

    The paper develops an In situ Fabricator prototype with integrated sensing, planning, control, architectural software, and interchangeable fabrication tools.

  • Results

    The IF1 demonstrators achieved millimetre-scale positioning for a full-size undulating brick wall and showed potential for enabling the Mesh Mould building process.

  • Takeaways & Limitations

    Mobile In situ Fabricators can connect digital design and construction through full-scale robotic fabrication and feedback-aware building processes.

  • Takeaways & Limitations

    The IF1 approach is constrained by the weight and control limitations of classical industrial arms, motivating a new actuator and the IF2 concept.

Abstract

from arXiv · show

This paper presents the concept of an In situ Fabricator, a mobile robot intended for on-site manufacturing, assembly and digital fabrication. We present an overview of a prototype system, its capabilities, and highlight the importance of high-performance control, estimation and planning algorithms for achieving desired construction goals. Next, we detail on two architectural application scenarios: first, building a full-size undulating brick wall, which required a number of repositioning and autonomous localisation manoeuvres. Second, the Mesh Mould concrete process, which shows that an In situ Fabricator in combination with an innovative digital fabrication tool can be used to enable completely novel building technologies. Subsequently, important limitations and disadvantages of our approach are discussed. Based on that, we identify the need for a new type of robotic actuator, which facilitates the design of novel full-scale construction robots. We provide brief insight into the development of this actuator and conclude the paper with an outlook on the next-generation In situ Fabricator, which is currently under development.

1 Introduction

The paper addresses the gap between digital design and largely manual on-site construction by proposing mobile robots for in situ fabrication. It frames robustness in changing, cluttered construction environments as the central challenge and introduces the In situ Fabricator concept.

  • On-site construction remains comparatively low in automation, with manual final assembly breaking the digital process chain.
  • In situ fabrication brings digitally controlled manufacturing and assembly directly to the construction site.
  • Mobile robots can fabricate structures larger than the machines themselves, avoiding the size constraint of large gantry systems.
  • Existing mobile construction robots were limited by standardised workflows, manual repositioning, or insufficient sensing and control.
  • Robust operation requires state estimation, control, and planning designed for unstructured, changing, dirty, and cluttered construction sites.
  • The paper presents a mobile In situ Fabricator class, an IF1 prototype, integrated digital planning and control, and two full-scale demonstrators.

2 Requirements and Definition of an In situ Fabricator

An In situ Fabricator is defined as a mobile construction robot combining fabrication versatility, site mobility, sensing, control, and integration with architectural planning. The concept excludes the entire building-production chain and focuses on producing the structure with feedback from the process.

  • In situ Fabricators are intended to support broad fabrication tasks through mobility, configurable end effectors, and operation in typical construction environments.
  • The robot must provide 1 to 5 millimetre end-effector positioning accuracy and operate with limited human intervention.
  • It must remain effective on non-flat terrain, around obstacles, and when moving humans or changing scenes lie outside its local operating area.
  • The definition includes practical constraints such as reaching standard-wall height, fitting through an 80 cm door, and being transportable by pallet or van.
  • Integration requirements include supplying robot and building-state information to architectural planning and control environments and supporting non-expert operators.
  • The scope excludes logistics and supply management while emphasising an in-the-loop feedback connection between design and building.

3 In situ Fabricator 1

IF1 is a first full-scale prototype built largely from off-the-shelf components, combining a 2.55 m, 40 kg industrial arm with mobile hardware, onboard power, sensing, computation, and tool interfaces. Its architecture supports configurable construction equipment but inherits known industrial-arm limitations.

  • The prototype was realised in 2014, partially based on Dimrob components, and mostly assembled from commercially available off-the-shelf parts.
  • IF1 uses an ABB IRB 4600 arm with 2.55 m reach and 40 kg payload mounted on a mobile base.
  • Its position-controlled arm accepts reference position and velocity commands through a manufacturer-provided interface at 250 Hz.
  • Four Li-Ion battery packs provide 3–4 hours of autonomous operation at average load without mains power.
  • An onboard hydraulic system powers the tracks and can also supply hydraulic power to end-effector tools.
  • The end-effector interface provides standard mounting, power, and data connections for a range of tools and temporary equipment.

4 State Estimation, Planning and Control

IF1 combines tailored sensing, CAD alignment, feedback, and optimal-control methods to localise and operate accurately during construction despite changing sites and model uncertainty.

  • Sensing and State Estimation: IF1’s sensing system combines robot localisation, CAD-model alignment, and feedback of building accuracy for precise on-site fabrication.The system is designed around fixed-reference localisation and construction-specific accuracy requirements.
  • Sensing and State Estimation: Initial laser-range sensing built registered 3D point clouds but assumed most surroundings remained unchanged, motivating localisation from nearby workpiece features.Changing construction environments made the original point-cloud strategy unsuitable as a general solution.
  • Sensing and State Estimation: Camera-based sensing supports both AprilTag-based robot localisation and stereo detection of the next wire for mesh fabrication and collision avoidance.The same camera model is reconfigured for distinct sensing tasks, highlighting its adaptability.
  • Feedback of Building Accuracy: Feedback measures the last welded wire in the global frame and adjusts subsequent mesh layers when deflection causes contour errors.This compensates for internal wire tension that is difficult to model accurately.
  • Planning and Control: Optimal control computes trajectories or feedback laws under constraints, enabling robust positioning despite model uncertainties and external perturbations.IF1 uses Constrained SLQ in an MPC fashion for repositioning while maintaining an end-effector position constraint.
  • Planning and Control: The approach handles moderate obstacle counts but not heavily cluttered environments with many intersecting or dynamically changing obstacles.The authors identify integration with higher-level planners as future work for such environments.

5 Integration into Architectural Design and Planning Software

The In situ Fabricator is integrated with architectural planning so designers can access robotic capabilities while fabrication data informs ongoing design and construction decisions.

  • Architectural Integration: Integrating IF capabilities into architectural planning aims to make robotic fabrication features directly available to architects and designers.The broader goal is for shape generation and rationalisation to reflect material, assembly, mobility, and workspace constraints.
  • Architectural Integration: Sensing feedback lets the design environment detect unforeseen assembly tolerances and process uncertainties during fabrication.This connects manipulation, sensing, and running design decisions rather than treating design as fixed before construction.
  • Architectural Integration: Grasshopper Rhinoceros implements high-level task planning, while a TCP/IP plugin provides online control and access to robot estimation, planning, and movement primitives.The integration covers mobile positions, fabrication sequencing, and arm, base, and end-effector commands.

6 Architectural Demonstrators and Examples

IF1 demonstrated semi-autonomous fabrication of an undulating brick wall and Mesh Mould elements in construction-like settings, using repositioning, localisation, sensing, and digital fabrication integration.

  • Undulating Brick Wall: The wall-building process divided construction into reachable brick patches executed from successive robot locations.The robot repeatedly moved, localised its base pose, and built each patch within its workspace.
  • Undulating Brick Wall: CAD-site alignment allowed the wall’s parametric geometry to be adapted to true construction-site dimensions before fabrication.The alignment used key environmental features derived from an initial 3D scan.
  • Undulating Brick Wall: The brick wall was successfully constructed semi-autonomously with 14 repositionings and a maximal assembly error of 7 mm.Localisation against an initial reference scan prevented global localisation and brick-placement errors from accumulating.
  • Undulating Brick Wall: The brick-laying operation was autonomous, but feeding bricks to the robot remained manual.
  • Mesh Mould: Mesh Mould combined IF1 with bespoke steel-mesh fabrication to create mould-and-reinforcement elements for customised reinforced-concrete structures.The system was tested by fabricating undulated, doubly curved Mesh Mould elements at NEST.
  • Mesh Mould: Mesh Mould’s integrated vision feedback enabled on-site correction for material tolerances and deformation during fabrication.The end-effector perception system supported global repositioning and local pose correction.

7 Limitations and lessons learned from IF1 - why classical industrial arms are a poor choice for mobile building construction robots

IF1’s industrial-arm design imposed weight, transport, safety, force-control, and durability drawbacks that limit mobile construction use and motivate new actuation approaches.

  • These shortcomings are presented as limitations of available off-the-shelf technology rather than defects specific to IF1.
  • At 1.4 tons, IF1 was too heavy for some standard building environments because industrial arms have low payload-to-weight ratios and require heavy bases.The ABB IRB 4600 example has a payload-to-weight ratio of 40 kg:440 kg.
  • Purely position-controlled arms are ill-suited to construction tasks requiring controlled interaction forces between tools and workpieces.A multi-DoF force-torque sensor can add flexibility but remains a sub-optimal workaround.
  • Classical electrically actuated arms without compliant, vibration-damping elements can suffer rapid gearbox damage and wear under poor construction load cases.The paper identifies series-elastic or hydraulic actuators as common solutions.
  • Platforms combining sufficiently sized hardware, joint-level force-torque control, and accessible low-level control loops were commercially unavailable.

8 Developing the next-generation In situ Fabricator

The next-generation IF2 requires a lighter, stronger, more controllable actuator and mobile platform than conventional systems provide, leading to an integrated hydraulic actuator and a legged-wheeled design.

  • IF2 requirements include agility, a 60 kg payload, a maximum 440 kg system weight, at least one 7-DOF arm with 2.5 m reach, safety modes, construction-site robustness, and high-bandwidth force control.
  • Conventional electrical or hydraulic joint actuators could not simultaneously meet IF2’s 440 kg system-mass limit and desired payload-to-weight ratio.The assessment therefore called for a novel actuator design combining performance, weight, and force-control properties.
  • The novel titanium vane actuator integrates hydraulic controls, safety valves, sensors, electronics, local processing, data buses, and slip rings in a lightweight package.Major components use laser powder-bed additive manufacturing to achieve compact structural integration.
  • The actuator is designed in three sizes and supports joint rotation around or perpendicular to its major axis for different manipulator segments.
  • Hydraulic actuation was selected because it offers high power density, scalability to construction-relevant dimensions, and robustness for mobile construction systems.
  • IF2’s preliminary design uses legs and wheels to support walking, driving, and hybrid locomotion modes.The prototype was under development, with first complete results expected by the end of 2017.

9 Summary and Conclusion

The paper defines In situ Fabricators as digitally controlled mobile robots for on-site manufacturing, assembly, and digital fabrication, and demonstrates the IF1 prototype through two full-scale applications. It also identifies limitations of the classical industrial robotics approach and motivates compact actuator-based designs for the next-generation IF2.

  • Concept and prototype: The paper presents In situ Fabricators as a class of robots whose requirements, prototype capabilities, and software algorithms are shaped by full-scale construction applications.The IF1 software framework used state estimation, motion planning, and control algorithms to meet desired accuracy and performance.
  • Full-scale demonstrators: IF1 achieved mm-scale positioning accuracy while building a full-size undulating brick wall through repeated repositioning manoeuvres.The system also successfully built metal mesh segments for the Mesh Mould process, which were filled with concrete and subjected to structural load tests.
  • Full-scale demonstrators: The Mesh Mould demonstrator showed that an In situ Fabricator combined with an innovative toolhead can enable novel building processes.
  • Limitations and next steps: The authors identify significant disadvantages in applying the classical industrial robotics approach to construction robotics and digital fabrication.These limitations motivated development of a new actuator concept.
  • Limitations and next steps: IF2 combines highly integrated compact actuators with efficient additively manufactured structural components as a proposed next-generation In situ Fabricator.The authors expect this development to advance full-scale construction robots.
Loading 1701.03573v1…