Swarm Of Autonomous Robots Research Articles

The Role of Time in a Robotic Swarm: A Joint View on Communications, Localization, and Sensing

Autonomous robotic swarms are envisioned for a variety of sensing applications in space exploration, search and rescue, and disaster management. An important capability of a swarm is sensing spatio-temporal processes such as radio wave propagation and seismic activities. The spa-tio-temporal properties of these processes dictate the required sensing position and time accuracy, as well as update rate. A dedicated wireless communication system needs to be jointly designed for swarm information exchange, self-localization, and sensing. In this article, we emphasize the role of time in a robotic swarm. We introduce the system ingredients and dive into realistic clock models. Clock models and channel access schemes decisively influence the swarm self-localization and synchronization accuracy, and consequently the swarm sensing performance. Finally, we discuss practical implementation aspects, introduce our developed swarm radio system, and give an outlook on a moon-analog exploration mission.

Decentralized Bioinspired Non-Discrete Model for Autonomous Swarm Aggregation Dynamics

In this paper a microscopic, non-discrete, mathematical model based on stigmergy for predicting the nodal aggregation dynamics of decentralized, autonomous robotic swarms is proposed. The model departs from conventional applications of stigmergy in bioinspired path-finding optimization, serving as a dynamic aggregation algorithm for nodes with limited or no ability to perform discrete logical operations, aiding in agent miniaturization. Time-continuous simulations were developed and carried out where nodal aggregation efficiency was evaluated using the following metrics: time to aggregation equilibrium, agent spatial distribution within aggregate (including average inter-nodal distance, center of mass of aggregate deviation from target), and deviation from target agent number. The system was optimized using cost minimization of the above factors through generating a random set of cost datapoints with varying initial conditions (number of aggregates, agents, field dimensions, and other specific agent parameters) where the best-fit scalar field was obtained using a random forest ensemble learning strategy and polynomial regression. The scalar cost field global minimum was obtained through basin-hopping with L-BFGS-B local minimization on the scalar fields obtained through both methods. The proposed optimized model describes the physical properties that non-digital agents must possess so that the proposed aggregation behavior emerges, in order to avoid discrete state algorithms aiming towards developing agents independent of digital components aiding to their miniaturization.

Bacterial quorum sensing applied to the coordination of autonomous robot swarms

This paper proposes a strategy for the coordination of a swarm of robots in an unknown environment. The basic idea is to achieve the autonomous movement of the group from an initial region to a target region avoiding obstacles. We use a behavior model similar to bacterial Quorum Sensing (QS) as a technique for the coordination of robots. This behavior has been described as a key element in the interaction between bacteria, and we use it as a basic tool for local interaction, both between the robot and between the robot and the environment. The movement of the swarm of robots, or multi-agent robotic system, is shown as an emerging behavior resulting from the interaction of agents (in the context of artificial intelligence) from basic rules of behavior. The proposed strategy was successfully evaluated by simulation on a set of robots.

A Dynamic Inertial Weight Strategy in Micro PSO for Swarm Robots

A relatively new area of research and development is Swarm Robotics. It is a part of the swarm intelligence field. In the proposed paper, we shall use swarm robotics in the field of defense and security, particularly for the problem of counter-improvised explosive device (IED) operations. The biggest problem in this regard is to physically detect the IEDs. We propose the use of a swarm of autonomous robots which shall be moving through the search space to collectively detect IEDs in a relatively lesser span of time with greater reliability. Since the robots are autonomous, there will not be any human contact involved, thus distancing humans from any potential IEDs or hazardous environments. The robot hardware shall be robust and able to traverse different kinds of terrains or even water bodies. A major problem of decision making for autonomous robots is localization of the robots towards the origin. Localization deals with finding its Cartesian coordinates and direction in the given coordinate system. For effective autonomous navigation of a robot, finding the position of the robot is essential at every point of time. Particle swarm optimization (PSO) is a useful method for population based global search. The proposed algorithm is an extension of micro-particle swarm optimization (µPSO) for Simultaneous Localization and Mapping. The effectiveness of this method is estimated by comparing its results with the traditional PSO and µPSO.

Pruning deep convolutional neural networks for efficient edge computing in condition assessment of infrastructures

AbstractHealth monitoring of civil infrastructures is a key application of Internet of things (IoT), while edge computing is an important component of IoT. In this context, swarms of autonomous inspection robots, which can replace current manual inspections, are examples of edge devices. Incorporation of pretrained deep learning algorithms into these robots for autonomous damage detection is a challenging problem since these devices are typically limited in computing and memory resources. This study introduces a solution based on network pruning using Taylor expansion to utilize pretrained deep convolutional neural networks for efficient edge computing and incorporation into inspection robots. Results from comprehensive experiments on two pretrained networks (i.e., VGG16 and ResNet18) and two types of prevalent surface defects (i.e., crack and corrosion) are presented and discussed in detail with respect to performance, memory demands, and the inference time for damage detection. It is shown that the proposed approach significantly enhances resource efficiency without decreasing damage detection performance.

Adaptive Online Fault Diagnosis in Autonomous Robot Swarms.

Previous work has shown that robot swarms are not always tolerant to the failure of individual robots, particularly those that have only partially failed and continue to contribute to collective behaviors. A case has been made for an active approach to fault tolerance in swarm robotic systems, whereby the swarm can identify and resolve faults that occur during operation. Existing approaches to active fault tolerance in swarms have so far omitted fault diagnosis, however we propose that diagnosis is a feature of active fault tolerance that is necessary if swarms are to obtain long-term autonomy. This paper presents a novel method for fault diagnosis that attempts to imitate some of the observed functions of natural immune system. The results of our simulated experiments show that our system is flexible, scalable, and improves swarm tolerance to various electro-mechanical faults in the cases examined.

A chunk-based slicer for cooperative 3D printing

PurposeThe purpose of this research is to develop a new slicing scheme for the emerging cooperative three-dimensional (3D) printing platform that has multiple mobile 3D printers working together on one print job.Design/methodology/approachBecause the traditional lay-based slicing scheme does not work for cooperative 3D printing, a chunk-based slicing scheme is proposed to split the print job into chunks so that different mobile printers can print different chunks simultaneously without interfering with each other.FindingsA chunk-based slicer is developed for two mobile 3D printers to work together cooperatively. A simulator environment is developed to validate the developed slicer, which shows the chunk-based slicer working effectively, and demonstrates the promise of cooperative 3D printing.Research limitations/implicationsFor simplicity, this research only considered the case of two mobile 3D printers working together. Future research is needed for a slicing and scheduling scheme that can work with thousands of mobile 3D printers.Practical implicationsThe research findings in this work demonstrate a new approach to 3D printing. By enabling multiple mobile 3D printers working together, the printing speed can be significantly increased and the printing capability (for multiple materials and multiple components) can be greatly enhanced.Social implicationsThe chunk-based slicing algorithm is critical to the success of cooperative 3D printing, which may enable an autonomous factory equipped with a swarm of autonomous mobile 3D printers and mobile robots for autonomous manufacturing and assembly.Originality/valueThis work presents a new approach to 3D printing. Instead of printing layer by layer, each mobile 3D printer will print one chunk at a time, which provides the much-needed scalability for 3D printing to print large-sized object and increase the printing speed. The chunk-based approach keeps the 3D printing local and avoids the large temperature gradient and associated internal stress as the size of the print increases.

Efficient collective swimming by harnessing vortices through deep reinforcement learning

Fish in schooling formations navigate complex flow fields replete with mechanical energy in the vortex wakes of their companions. Their schooling behavior has been associated with evolutionary advantages including energy savings, yet the underlying physical mechanisms remain unknown. We show that fish can improve their sustained propulsive efficiency by placing themselves in appropriate locations in the wake of other swimmers and intercepting judiciously their shed vortices. This swimming strategy leads to collective energy savings and is revealed through a combination of high-fidelity flow simulations with a deep reinforcement learning (RL) algorithm. The RL algorithm relies on a policy defined by deep, recurrent neural nets, with long-short-term memory cells, that are essential for capturing the unsteadiness of the two-way interactions between the fish and the vortical flow field. Surprisingly, we find that swimming in-line with a leader is not associated with energetic benefits for the follower. Instead, "smart swimmer(s)" place themselves at off-center positions, with respect to the axis of the leader(s) and deform their body to synchronize with the momentum of the oncoming vortices, thus enhancing their swimming efficiency at no cost to the leader(s). The results confirm that fish may harvest energy deposited in vortices and support the conjecture that swimming in formation is energetically advantageous. Moreover, this study demonstrates that deep RL can produce navigation algorithms for complex unsteady and vortical flow fields, with promising implications for energy savings in autonomous robotic swarms.

Motion Localization with Optic Flow for Autonomous Robot Teams and Swarms

The ability to localize moving objects within the environment is critical for autonomous robotic systems. This paper describes a moving object detection and localization system using multiple robots equipped with inexpensive optic flow sensors. We demonstrate an architecture capable of detecting motion along a plane by collecting three sets of one-dimensional optic flow data. The detected object is then localized with respect to each of the robots in the system.

Navigation of carlike robots in an extended dynamic environment with swarm avoidance

SummaryIn this paper, we propose a new solution to the motion planning and control problem for a team of carlike mobile robots traversing in an extended dynamic environment. Motivated by the emerging necessity to avoid or defend against a swarm of autonomous robots, the wide array of obstacles in this dynamic environment for the first time includes a swarm of boids governed separately by a system of ordinary differential equations. The swarm exhibits collective emergent behaviors, whereas the carlike mobile robots safely navigate to designated targets. We present a set of nonlinear continuous controllers for obstacle, collision, and swarm avoidance. The controllers provide a collision‐free trajectory within a constrained workspace cluttered with various fixed and moving obstacles while satisfying the nonholonomic and kinodynamic constraints associated with the vehicular robotic system. An advantage of the proposed method is the ease in deriving the acceleration‐based control laws from the Lyapunov‐based control scheme. The effectiveness of the control laws is demonstrated via computer simulations. The novelty of this paper lies in the simplicity of the controllers and the ease in the treatment of an extended dynamic environment, which includes swarm avoidance.

Pheeno, A Versatile Swarm Robotic Research and Education Platform

Swarms of low-cost autonomous robots can potentially be used to collectively perform tasks over very large domains and time scales. Novel robots for swarm applications are currently being developed as a result of recent advances in sensing, actuation, processing, power, and manufacturing. These platforms can be used by researchers to conduct experiments with robot collectives and by educators to include robotic hardware in their curricula. However, existing low-cost robots are specialized and can lack desired sensing, navigation, control, and manipulation capabilities. This letter presents a new mobile robot platform, Pheeno, that is affordable, versatile, and suitable for multirobot research, education, and outreach activities. Users can modify Pheeno for their applications by designing custom modules that attach to its core module. We describe the design of the Pheeno core and a three degree-of-freedom gripper module, which enables unprecedented manipulation capabilities for a robot of Pheeno's size and cost. We experimentally demonstrate Pheeno's ability to fuse measurements from its onboard odometry for global position estimation and use its camera for object identification in real time. We also show that groups of two and three Pheenos can act on commands from a central controller and consistently transport a payload in a desired direction.

Division of labor in a swarm of autonomous underwater robots by improved partitioning social inhibition

In this paper, a distributed algorithm for adaptive task allocation and adaptable partitioning of a swarm into different work-groups is proposed and used in a swarm of underwater robots. The algorithm is based on local interactions of agents and is inspired by honeybee age-polyethism. It is adaptive to changes in the swarm size (workforce) and relative demands (workload) for different tasks and it limits the number of switchings of agents between different tasks enabling specialization of agents. The preliminary version of the algorithm was introduced previously. Here a fully decentralized version of the algorithm is proposed that improves the previous version by removing the need for global information. The algorithm is successfully implemented in swarms of physically-embodied underwater robots while the swarm size and the demands for the tasks change over the course of the experiments.

Artificial Swarm Intelligence, a Human-in-the-Loop Approach to A.I.

Most research into Swarm Intelligence explores swarms of autonomous robots or simulated agents. Little work, however, has been done on swarms of networked humans. This paper introduces UNU, an online platform that enables networked users to assemble in real-time swarms and tackle problems as an Artificial Swarm Intelligence (ASI). Modeled after biological swarms, UNU enables large groups of networked users to work together in real-time synchrony, forging a unified dynamic system that can quickly answer questions and make decisions. Early testing suggests that human swarming has significant potential for harnessing the Collective Intelligence (CI) of online groups, often exceeding the natural abilities of individual participants.

Success Rate of Creatures Crossing a Highway as a Function of Model Parameters

In modeling swarms of autonomous robots, individual robots may be identified as cognitive agents. We describe a model of population of simple cognitive agents, naïve creatures, learning to safely cross a cellular automaton based highway. These creatures have the ability to learn from each other by evaluating if creatures in the past were successful in crossing the highway for their current situation. The creatures use “observational social learning” mechanism in their decision to cross the highway or not. The model parameters heavily influence the learning outcomes examined through the collected simulation metrics. We study how these parameters, in particular the knowledge base, influence the creatures’ success rate of crossing the highway.

Determination of Spatial Configuration of an Underwater Swarm with Minimum Data

This paper is the extension of work presented at the IARP Conference “Bio inspired robotics” held in Frascati (Italy), 14 May 2014. The subject is the localization problem of an underwater swarm of autonomous underwater robots (AUV), in the frame of the HARNESS project; by localization, we mean the relative swarm configuration, i.e., the geometrical shape of the group. The result is achieved by using the signals that the robots exchange. The swarm is organized by rules and conceived to perform tasks, ranging from environmental monitoring to terrorism-attack surveillance. Two methods of determining the shape of the swarm, both based on trilateration calculation, are proposed. The first method focuses on the robot's speed. In this case, we use our knowledge of the speeds and distances between the machines, while the second method considers only distances and the orientation angles of the robots. Unlike a trilateration problem, we do not know the position of the beacons and this renders the problem a difficult one. Moreover, we have very few data. More than one step of motion is needed to resolve the multiple solutions found, owing to the symmetries of the system and optimization process of one or more objective functions leading to the final configuration. We subsequently checked our algorithm using a simulator taking into account random errors affecting the measurements.

Costs and benefits of behavioral specialization

In this work, we study behavioral specialization in a swarm of autonomous robots. In the studied swarm, robots have to carry out tasks of different types that appear stochastically in time and space in a given environment. We consider a setting in which a robot working repeatedly on tasks of the same type improves its performance on them due to learning. Robots can exploit learning by adapting their task selection behavior, that is, by selecting with higher probability tasks of the type on which they have improved their performance. This adaptation of behavior is called behavioral specialization. We employ a simple task allocation strategy that allows a swarm of robots to behaviorally specialize. We study the influence of different environmental parameters on the performance of the swarm and show that the swarm can exploit learning successfully. However, there is a trade-off between the benefits and the costs of specialization. We study this trade-off in multiple experiments using different swarm sizes. Our experimental results indicate that spatiality has a major influence on the costs and benefits of specialization.

Autonomy Spectrum and Performance Perception Issues in Swarm Supervisory Control

This paper proposes a two-part contribution to the human factor issues raised by the prospect of the deployment of swarms of autonomous robots under the supervision of a human operator. First, we propose a generic framework for system description that allows us to analyze and guide the design of automated systems. Second, we analyze the way some highly adaptive systems such as swarms can be perceived by operators, and especially the way representational gaps between man and machine can impact global performances.

A Distributed Algorithm for Un-balanced Partitioning of a Swarm of Autonomous Mobile Robots and Its Performance Analysis

This piece of work studies the partitioning problem on independently operating swarm of autonomous mobile robots and devises algorithms for unbalanced partitioning in a distributed computing environment. The robots considered here are all identical and are very simple and weak. There is no central control over the robots and the robots do not communicate among themselves. Each robot executes the same algorithm based on their local information. This paper frames the algorithms for unbalanced partitioning by sorting the robots based on their ranking and then allocating them in different groups based on their ranks, such that N robots are divided into K unbalanced groups of unequal robots in each group. This paper also presents the performance based analysis of the un-balanced algorithm U_PART over the balanced algorithm and examines their effects via different examples through 50 separate test cases. It also tries to bring out the shortcomings of the proposed U_PART algorithm and proposes another alternative approach towards un-balanced partitioning to overcome the limitation of the U_PART algorithm.

Biologically inspired collective comparisons by robotic swarms

Intelligent entities must often make decisions by comparing several candidate alternatives and selecting the best one. This is just as true for autonomous swarms as it is for solitary robots, but to date there has been little work to propose efficient comparison behaviors for autonomous robotic swarms that are not tied to specific environments. In this work, we examine an elegant collective comparison strategy that is used by at least three different species of social insect and adapt it for artificial systems. The behavior is particularly attractive for robotic implementations because it relies only on short range explicit peer-to-peer communication, eliminating the need for chemical trails or other forms of stigmergy. The proposed comparison strategy is proven to converge, and a series of experiments using real robots with noisy sensors is presented that validates our theoretical analysis. Using the proposed behavior, a robotic swarm is able to compare alternatives collectively more accurately than its member robots would be able to individually.

Towards autonomous robot swarms for multi-target localisation and monitoring with applications to counter IED operations

A swarm robotics approach is adopted in designing a fully autonomous multi-robot-based solution to the problem of locating generic targets within a given search space. A proof of concept system is developed and tested within a 3D simulation environment. A series of laboratory experiments are carried out to assess the performance of the system with respect to the given task of localising and monitoring generic targets, with reference to a counter IED scenario. Further experiments are carried out to evaluate the robustness to robot failure and scalability of the system.