Conference PaperPDF Available

Autonomous Adaptive Modification of Unstructured Environments

Authors:

Abstract and Figures

We present and validate a property-driven autonomous system that modifies its environment to achieve and maintain navigability over a highly irregular 3-dimensional terrain. In our approach we use decision procedures that tie building actions to the terrain model, giving rise to adaptive and robust building behavior. The building algorithm is driven by continuous evaluation and reaction to terrain properties, rather than relying on a structure blueprint. This capability is essential in robotic systems that operate in unstructured outdoor or remote environments, either on their own or as part of a team. We demonstrate the effectiveness of our approach by running a low-cost robot system that can build with compliant bags in a variety of irregular terrains.
Content may be subject to copyright.
A preview of the PDF is not available
... Other projects have demonstrated spraying foams to smooth rough terrain at a small scale, a means of site preparation which could, e.g., build ramps conforming to unstructured terrains using amorphous materials [54]. Similar operations could be achieved at a larger scale with the Digital Construction Platform, a mobile robot capable of both excavation and depositing structural foam (Fig. 12A) [55]. ...
... Automating the dry-stacking of stones as a method of foundation support has recently been the subject of academic attention [69,70]. Other laboratory demonstrations have focused on the ability to autonomously stack small sand bags [54,71]. However, industry processes remain heavily dependent on concrete foundations established using heavy machinery. ...
Article
Real-world construction projects typically require three groups of tasks: site preparation (earthmoving, leveling), substructure (anchoring, foundations), and superstructure (load-bearing elements, facade, plumbing, wiring, etc.). Advances in construction automation have revealed a gap between industry and academic research, where industry efforts have been focused on automating conventional earthmoving equipment and embracing prefabrication in order to reduce the amount of work that needs to be done on site, while academic efforts have largely concentrated on proposals for on-site additive manufacturing or discrete assembly, which may be of limited applicability to industry. This review presents a broad range of advancements in construction automation research, and finds that achieving fully autonomous construction in unstructured environments will require considerably more development in all three groups of construction tasks, as well as a particular emphasis on coordinating myriad construction tasks between different task-specific robots. Consideration is given to both mature technologies (conventional equipment widely used in industry) and emerging technologies (novel machines designed for autonomy). Key findings from the survey suggest that achieving the goal of fully autonomous construction will require more attention to be paid to site preparation and substructure tasks, material-robot systems (co-designed robots and building materials), embedded sensing, auxiliary construction tasks, and coordinating operations between robot systems. More general lessons from the literature indicate that making incremental improvements to mature technologies may benefit the industry in the short term, but there are considerable limitations to adding autonomy to equipment designed for human operators. Instead, we perceive a demand for novel hardware to be developed for specific tasks, in each case based on fundamental principles and at the appropriate scale, as well as for an increase in interdisciplinary research. We suggest that the reported shortage of skilled labor in the industry can be met with an increased emphasis on training for leveraging advances in automation.
... It has received wide robot explorers to use robotics for a number of scenarios by facilitating the potential development of the web based robotic ecosystem. The uFactory [30] is specialized in developing and manufacturing consumer robotics systems aiming to make people believe that humanity is going to benefit from robots in our daily life, and that they will become a necessary item for everybody in the future. Moharana et al. [31] performed a high speed-based laser cutting based operation with parallel delta robot which gave a precise and satisfactory result in comparison to serial robots. ...
Article
Full-text available
Laser engraving is the most non-traditional and efficient working method in the machining of materials of different geometry as compared to conventional methods. The main objective of this study is to determine the impact of uArm swift pro robot operated laser engraving process on a wooden pitch board piece. However, the robot was connected with uArm Studio 1.1.22 software to perform laser engraving operation. For this purpose the effect of process parameters like spot diameter and depth of penetration were investigated with different working length of the robot end effector, measured from wooden pitch board base. Experimental observation method was used to investigate the formation of deep and light engraving pattern on the pitch board surface by measuring penetration depth and spot diameter in suitable condition. The result obtained from the experiment and statistical parameters showed a new dimension to find a suitable working length of the robot assisted laser nozzle where the laser penetration effect was clearly perceptible for the wooden material.
... Despite the recent advances in the context of on-site digital fabrication, most applications only consider the use of regular materials, such as prefabricated bricks (Dörfler et al., 2016), and there still exist only a few robots integrating all the aforementioned capabilities. For example, in Saboia et al. (2018) and Fujisawa et al. (2015), completely autonomous systems are shown capable of constructing auxiliary structures to achieve and maintain navigability across previously untraversable terrain. However, they use customized compliant bags as construction material or apply polyurethane foam, respectively, and not naturally occurring building materials such as stones. ...
Article
Automating building processes through robotic systems has the potential to address the need for safer, more efficient, and sustainable construction operations. While ongoing research effort often targets the use of prefabricated materials in controlled environments, here we focus on utilizing objects found on‐site, such as irregularly shaped rocks and rubble, as a way of enabling novel types of construction in remote and extreme environments, where standard building materials might not be easily accessible. In this article, we present a perception and grasp pose planning pipeline for autonomous manipulation of objects of interest with a robotic walking excavator. The system incrementally builds a LiDAR‐based map of the robot's surroundings and provides the ability to register externally reconstructed point clouds of the scene, for example, from images captured by a drone‐borne camera, which helps increasing map coverage. In addition, object‐like instances, such as stones, are segmented out of this map. Based on this information, collision‐free grasping poses for the robotic manipulator are planned to enable picking and placing of these objects, while keeping track of them during the manipulation. The approach is validated in a real setting on an architectural relevant scale by segmenting and manipulating boulders of several hundred kilograms, which is a first step towards the full automation of dry‐stack wall building processes. Video – https://youtu.be/4bc5n2-zj3Q
... Further, Tosun et al. (2018) have looked at planning for construction of functional structures by modular robots, focusing on identifying features that enable environment modification in order to make the terrain traversable. In similar work, Saboia et al. (2018) have looked at modification of unstructured environments using objects, to create ramps that enhance navigability. More recently, Choi et al. (2018) extended the cognitive architecture ICARUS to support the creation and use of functional structures such as ramps, in abstract planning scenarios. ...
Article
Full-text available
Robots in the real world should be able to adapt to unforeseen circumstances. Particularly in the context of tool use, robots may not have access to the tools they need for completing a task. In this paper, we focus on the problem of tool construction in the context of task planning. We seek to enable robots to construct replacements for missing tools using available objects, in order to complete the given task. We introduce the Feature Guided Search (FGS) algorithm that enables the application of existing heuristic search approaches in the context of task planning, to perform tool construction efficiently. FGS accounts for physical attributes of objects (e.g., shape, material) during the search for a valid task plan. Our results demonstrate that FGS significantly reduces the search effort over standard heuristic search approaches by ≈93% for tool construction.
... A few systems based on 3D-printing large-scale ceramic structures have been demonstrated outside controlled environments, though still on prepared flat surfaces [3,4]. Some work has considered adapting to uneven terrains by depositing amorphous materials [5,6], building truss structures to conform to terrains [7], or actuating structural elements to conform to new site conditions in real-time [8]. Autonomous site preparation and substructure tasks have been largely absent from academic research, though work on automating conventional earthmoving equipment has been proposed [9]. ...
Article
Advances in construction automation have tended to focus on either automating conventional earthmoving equipment or on the discrete assembly of superstructure elements. Neither paradigm has addressed anchoring introduced material into soil, a critical task for virtually all useful structures. Simple anchoring can be achieved by driving posts (discrete linear elements) or sheet piles (interlocking profiles) into the ground, serving as a foundation for a later superstructure. In this paper we present Romu, a wheeled robot that uses a combination of a vibratory hammer and its own body mass to drive both posts and piles into the ground. We report on the effects of hardware parameters on pile driving performance, and demonstrate operation in both controlled and natural environments. Romu is first configured to drive interlocking sheet piles. In addition to their utility as foundations, such walls could be useful directly as check dams, interventions used to prevent erosion and promote groundwater recharge in arid regions. We use simulations based on real-world terrains to explore the potential impact of a fleet of such robots deployed over a large watershed region, using a simple reactive approach to dynamically determine dam placement. Romu is then configured to drive a range of readily available building materials that commonly serve as posts. These include wooden slats that can be used for sand fencing, an intervention used to collect wind-blown sand to build barrier dunes. Post driving performance is characterized for a range of materials, and finally the use case of sand fencing is evaluated using physical tests at 1:10 scale in order to predict its potential impact. To broaden the utility of such robots in field settings, directions for future work include refinement of the hardware for improved operation in more terrains, increased capabilities for fuller autonomy, and integration with other construction tasks for more complex projects.
... Further, [17] has looked at planning for construction of functional structures by modular robots, focusing on identifying features that enable environmental modification in order to make it traversable. In similar work, [18] has looked at modification of unstructured environments using objects, to create ramps that enhance navigability. More recently, [19] extended the cognitive architecture ICARUS to support the creation and use of functional structures such as ramps, in abstract planning scenarios. ...
Preprint
Macgyvering refers to solving problems inventively by using whatever objects are available at hand. Tool Macgyvering is a subset of macgyvering tasks involving a missing tool that is either substituted (tool substitution) or constructed (tool construction), from available objects. In this paper, we introduce a novel Tool Macgyvering framework that combines tool substitution and construction using arbitration that decides between the two options to output a final macgyvering solution. Our tool construction approach reasons about the shape, material, and different ways of attaching objects to construct a desired tool. We further develop value functions that enable the robot to effectively arbitrate between substitution and construction. Our results show that our tool construction approach is able to successfully construct working tools with an accuracy of 96.67%, and our arbitration strategy successfully chooses between substitution and construction with an accuracy of 83.33%.
... Further, Tosun et al. have looked at planning for construction of functional structures by modular robots, focusing on identifying features that enable environment modification in order to make the terrain traversable [37]. In similar work, Saboia et al. have looked at modification of unstructured environments using objects, to create ramps that enhance navigability [28]. More recently, Choi et al. extended the cognitive architecture ICARUS to support the creation and use of functional structures such as ramps, in abstract planning scenarios [8]. ...
Preprint
Robots in the real world should be able to adapt to unforeseen circumstances. Particularly in the context of tool use, robots may not have access to the tools they need for completing a task. In this paper, we focus on the problem of tool construction in the context of task planning. We seek to enable robots to construct replacements for missing tools using available objects, in order to complete the given task. We introduce the Feature Guided Search (FGS) algorithm that enables the application of existing heuristic search approaches in the context of task planning, to perform tool construction efficiently. FGS accounts for physical attributes of objects (e.g., shape, material) during the search for a valid task plan. Our results demonstrate that FGS significantly reduces the search effort over standard heuristic search approaches by approximately 93% for tool construction.
Chapter
The ability to build structures with autonomous robots using only found, minimally processed stones would be immensely useful, especially in remote areas. Assembly planning for dry-stacked structures, however, is difficult since both the state and action spaces are continuous, and stability is strongly affected by complex friction and contact constraints. We propose a planning algorithm for such assemblies that uses a physics simulator to find a small set of feasible poses and then evaluates them using a hierarchical filter. We carefully designed the heuristics for the filters to match our goal of building stable, free-standing walls. These plans are then executed open-loop with a robotic arm equipped with a wrist RGB-D camera. Experimental results show that the proposed planning algorithm can significantly improve the state of the art in robotic dry stacking.
Article
Full-text available
Field robots are widely used to accomplish a variety of tasks in many different fields. However, setting of the locomotive ability of these robots at the design phase may prevent the traversal of unknown rough terrain. To address this shortcoming of existing robots, we designed a robot that is able to modify its environment by using polyurethane foam to construct auxiliary structures to facilitate movement across previously impassable terrain. Two robots were implemented with the ability to eject one-and two-part polyurethane foam, respectively. First, we investigated the specifications of the different types of polyurethane foam, specifically the volume expansion and curing time thereof. Two-part polyurethane foam cures in approximately 2 min, compared with 1 h for the one-part foam, but requires more accurate spraying, and its vertical expansion needs to be considered for accurate construction of auxiliary structures. The performance of each robot was tested in two experiments in the field. The first involved filling a deep ditch before crossing over it, while in the second experiment, each robot constructed a slope leading up to a high step, allowing the robot to move onto the step. Both robots succeeded in completing these tasks successfully, with the main difference in performance being the time taken before the robot was able to traverse the obstacles. Using two-part polyurethane foam resulted in much shorter curing times, although the structures constructed were not as even as those for the one-part polyurethane foam, and the robot needed to wait 10 s between the applications of each successive layer of foam to account for the vertical expansion of the material. Our findings demonstrate the effectiveness of our polyurethane foam construction robots in overcoming obstacles in unknown rough terrain.
Article
Full-text available
This paper presents a series of experiments where a group of mobile robots gather 81 randomly distributed objects and cluster them into one pile. Coordination of the agents' movements is achieved through stigmergy. This principle, originally developed for the description of termite building behaviour, allows indirect communication between agents through sensing and modification of the local environment which determines the agents' behaviour. The efficiency of the work was measured for groups of one to five robots working together. Group size is a critical factor. The mean time to accomplish the task decreases for one, two, and three robots respectively, then increases again for groups of four and five agents, due to an exponential increase in the number of interactions between robots which are time consuming and may eventually result in the destruction of existing clusters. We compare our results with those reported by Deneubourg et al. (1990) where similar clusters are observed in ant colonies, generated by the probabilistic behaviour of workers.
Article
We present hardware, perception, and planning tools that enable a modular robot to autonomously deploy passive structures to reach otherwise-inaccessible regions of the environment to accomplish high-level tasks. An environment characterization algorithm identifies features in the environment that can be augmented to create a path between two disconnected regions of the environment. Specially-designed building blocks enable the robot to create structures that can augment the identified features to expand the reachable space of the robot. A high-level planner reasons about the high-level task, robot locomotion capabilities, and the environment to decide if and where to augment the environment in order to perform the desired task. These autonomous, perception-driven augmentation tools extend the adaptive capabilities of modular robot systems.
Conference Paper
Current implementations of decentralized multi-robot construction systems are limited to construction of rudimentary structures such as walls and clusters, or rely on the use of blueprints for regulation. Building processes that make use of blueprints are unattractive in unknown environments as they can not compensate for heterogeneities, such as irregular terrain. In nature, social insects coordinate the construction of their nests using stigmergy, a mechanism of indirect coordination that is robust and adaptive. In this paper, we propose the design of a multi-robot construction platform called the Swarm Robotics Construction System (SRoCS). The SRoCS platform is designed to leverage stigmergy in order to coordinate multi-robot construction in unknown environments.
Article
In this paper, we develop an autonomous construction system in which self-contained ground robots build a protective barrier by means of compliant pockets. We present a stochastic control algorithm based on two biological mechanisms–stigmergy and templates–that takes advantage of compliant pockets for autonomous construction with single and multiple robots. The control algorithm guides the robot(s) to build the protective barrier without relying on a central planner, an external computer, or a motion capture system. We propose a statistical model to represent the structures built with the compliant pockets, and we provide a set of criteria for assessing the performance of the proposed system. To demonstrate the feasibility of the proposed system, real-robot and simulation experiments were carried out. The results show the viability of the proposed autonomous construction system.
Article
We present a model of construction using iterative amorphous depositions and give a distributed algorithm to reliably build ramps in unstructured environments. The relatively simple local strategy for interacting with irregularly shaped, partially built structures gives rise to robust adaptive global properties. We illustrate the algorithm in both single robot and multi-robot cases via simulations and describe how to solve key technical challenges to implementing this algorithm via a robotic prototype.
Conference Paper
We present and compare three different amorphous materials for robotic construction. By conforming to surfaces they are deposited on, such materials allow robots to reliably construct in unstructured terrain. However, using amorphous materials presents a challenge to robotic manipulation. We demonstrate how deposition of each material can be automated and compare their material properties, cost, and cost in time in order to evaluate their suitability for developing amorphous robotic construction system.
Article
Complex systems are characterized by many independent components whose low-level actions produce collective high-level results. Predicting high-level results given low-level rules is a key open challenge; the inverse problem, finding low-level rules that give specific outcomes, is in general still less understood. We present a multi-agent construction system inspired by mound-building termites, solving such an inverse problem. A user specifies a desired structure, and the system automatically generates low-level rules for independent climbing robots that guarantee production of that structure. Robots use only local sensing and coordinate their activity via the shared environment. We demonstrate the approach via a physical realization with three autonomous climbing robots limited to onboard sensing. This work advances the aim of engineering complex systems that achieve specific human-designed goals.