Robotic Swarm Intelligence for Lunar Exploration

Posted by in Science & Technology

Deep space presents numerous hazards and harsh conditions for remote exploration missions, which must often operate autonomously without intervention from Earth. To increase the survivability of remote missions, space agencies are exploiting principles and techniques that can help such systems become more resilient through self-management and automatic adaptation. By adhering to the principles of autonomic computing, contemporary spacecraft systems implement vital features for unmanned missions, such as self-configuration, self-healing, self-optimization, and self-protection.

Moreover, biologically inspired approaches target new classes of space exploration missions that use swarm intelligence and swarm cooperation to achieve extremely robust systems. Swarm-based systems comprise thousands of small spacecraft working together to explore places in deep space where a single and monolith spacecraft is impractical. However, developing such systems — from conceptualization to validation — is a complex multidisciplinary activity, and reliability and safety are key objectives. The systems can’t exhibit post-release faults or failures that could jeopardize the mission or cause loss of life. They integrate complex hardware and sophisticated software and thus require careful design and thorough testing to ensure adequate reliability. Moreover, aerospace systems have strict dependability and real-time requirements; need flexible resource reallocation; and must be limited in size, weight, and power consumption.

In this article, I discuss swarm intelligence and its applications in the space industry, particularly lunar and planetary missions. We discuss in detail the taxonomy of swarm behaviors and also discuss some important steps to be included while designing such systems for real-world challenging environments. In the end, I give a review of the state-of-the-art swarm robotic applications from research and industrial domains.

Swarm Intelligence (SI)

Swarm intelligence (SI) is the collective behavior of decentralized, self-organized systems, natural or artificial. SI systems are typically made up of a population of simple agents interacting locally with one another and with their environment. The inspiration often comes from nature, especially biological systems. The swarm agents follow very simple rules, and although there is no centralized control structure dictating how these individual agents should behave. The local and random interactions between such agents lead to the emergence of “intelligent” global behavior, unknown to the individual agents. Natural examples of SI include ant colonies, bird flocking, animal herding, bacterial growth, and fish schooling.

Importance of SI in Space Applications

Research in Swarm Intelligence (SI) started in the late 1980s. Besides the applications to conventional optimization problems, SI can be employed in a ton of scenarios like library materials acquisition, communications, medical dataset classification, dynamic control, heating system planning, moving objects tracking, and prediction. Indeed, SI can be applied to a variety of fields in fundamental research, engineering, industries, and social sciences. Nowadays, NASA is investigating the application of swarm computing to both spacecraft and surface-based rovers. The Autonomous NanoTechnology Swarm (ANTS) concept mission is a collaboration between NASA Goddard Space Flight Center and NASA Langley Research Center. It aims to develop revolutionary mission architectures and exploit AI techniques and paradigms in future space exploration.

A number of features of swarm intelligence are certainly attractive to the space engineering community. The space environment typically puts stringent constraints on the capabilities of single satellites, robots, or anything that needs to survive in space (space agents). Space agents are particularly limited in terms of mobility (propellant and power-limited), communication (power-limited), and size (mass-limited). At the same time, a high level of adaptability, robustness, and autonomy is required to increase the chances of success of operating in a largely unknown environment. Similar characteristics are found in the individual components of a biological swarm. Moreover, a number of space applications are naturally based on the presence of multiple space agents.

Swarm intelligence methods would represent an attractive design option allowing, for example, the achievement of autonomous operation of formations. Simple agents interacting locally could be considered as a resource rather than overhead. At the same time, one would be able to engineer systems that are robust, autonomous, adaptable, distributed, and inherently redundant. Besides, swarms allow for mass production of single components, thus promising mission cost reduction, and represent smaller, lighter weight highly stowable systems, thus allowing reduced launch costs. Recently, these motivations led a number of researchers to simulate some degree of swarm intelligence in a number of space systems and to investigate their behavior.

Multi-Agent and Swarm Intelligence in Lunar and Planetary Environments

Modern robotic platforms for in-situ space exploration have thus far involved single-robots equipped with a number of specialized sensors providing scientists with unique information about a planet’s surface. However, there is a number of exploration problems where larger spatial apertures of the exploration system are necessary, requiring a completely new perspective on in-situ space exploration and its required technologies. Large networks of swarm robots pave the way: agents in a swarm span ad-hoc communication networks, localize themselves based on radio signals, share resources, process data and make inferences over the network in a decentralized fashion. By cooperation, local information collected by agents becomes globally available.

Although the teleoperation of lunar rovers is possible from Earth and has been demonstrated successfully with the Lunokhod 1 and 2 rover missions, operator fatigue is of concern, especially when coordinating actions with teams of robots over extended missions. In addition, these systems must be robust to communication interruptions and bandwidth and latency issues. An alternative approach is the deployment of autonomous systems that require limited human control. There currently exist two major approaches to developing autonomous control systems: human knowledge/model-based controllers and systems based on machine learning techniques. Human knowledge/model-based behavior control strategies rely on human input in the form of ad-hoc control rules, task-specific assumptions, and human experience and knowledge. In contrast, machine learning systems of the type examined here perform task decomposition through ‘emergent’ self-organized behavior. In lunar and planetary environments, task-specific assumptions may not always be valid in-situ and may require reassessment during a mission. There is also a growing necessity for the development of generic teams of “utility” robots that can facilitate in-situ resource utilization and perform specific tasks that may never have been envisioned during mission planning and modeling stages.

Groups of multiple robots can collect spatially and temporally correlated data quickly and effectively, providing scientists with high-quality information on surface and sub-surface geology and chemistry, airborne particulates (includes dust, dirt, soot, smoke, and liquid droplets emitted into the air, is small enough to be suspended in the atmosphere), and weather patterns. A number of multi-agent and multi-sensor missions for planetary explorations have been proposed, including:

Advantages of Multi-robotic Approaches

  • Improved performance: if tasks can be decomposable then by using parallelism (i.e. concurrent execution of the different task on multiple computing nodes), groups can make tasks to be performed more efficiently.
  • Task enablement: groups of robots can do certain tasks that are impossible for a single robot.
  • Distributed sensing: the range of sensing of a group of robots is wider than the range of a single robot.
  • Distributed action: a group of robots can actuate in different places at the same time.
  • Fault tolerance: under certain conditions, the failure of a single robot within a group does not imply that the given task cannot be accomplished, thanks to the redundancy of the system.

Basic Swarm Behaviors For Swarm Robotics

In swarm robotics, multiple robots — homogeneous or heterogeneous — are interconnected, forming a swarm of robots. Since individual robots have processing, communication, and sensing capabilities locally on-board they are able to interact with each other and react to the environment autonomously.

In most swarm algorithms, individuals perform according to local rules and the overall behavior emerges organically from the interplay of the individuals of the swarm. Translated to the swarm robotics domain, individual robots exhibit a behavior that is based on a local rule set which can range from a simple reactive mapping between sensor inputs and actuator outputs to elaborate local algorithms. Typically, these local behaviors incorporate interactions with the physical world, including the environment and other robots. Each interaction consists of reading and interpreting the sensory data, processing this data, and driving the actuators accordingly. Such a sequence of interactions is defined as basic behavior that is repeatedly executed, either indefinitely or until a desired state is reached.


A swarm robotic system must exhibit several properties that are shown by natural swarms:

  • Flexibility

Swarm robotics aims to attain a variety of tasks. Here comes the feature of flexibility in focus. For the tasks, the system must be able to create various solutions by coordination and cooperation between robots. So, robots should find solutions by working together and be able to change their roles according to the given tasks. They should be capable of acting simultaneously according to the changes in their environment.

  • Scalability

Scalability means that the systems must be able to work with different sizes of groups. There should not be a global number of robots present in a swarm, but the sizes may differ, and accomplishing the task should still be possible and effective. The number of group members must not influence the performance of the system. So, swarm robotic systems should be able to operate with a different number of members. The system should work effectively when the swarm size is small and it should support coordination and cooperation amongst the members if the swarm size is large.

  • Robustness

A system is referred to as robust if it has the ability to continue operating even if there are environmental disturbances or system faults. Environmental disturbances may include the changing of the surroundings, addition in the number of obstacles in the environment, weather changes, and so on. Some of the system members can have a malfunction or can fail to perform. A swarm robotic system must be able to cope with such circumstances. In swarm robotic systems, individual robots are mainly very simple. This means that they cannot perform any significant tasks alone. So, if a system loses some robots it should not affect the overall performance of the system. The loss of individuals can be compensated by another member and the tasks must go on with the same level of efficiency.

The Taxonomy of Swarm Behaviors

Spatial Organization

These behaviors allow the movement of the robots in a swarm in the environment in order to spatially organize themselves or objects.

  • Aggregation moves the individual robots to congregate spatially in a specific region of the environment. This allows individuals of the swarm to get spatially close to each other for further interaction.
  • Object clustering and assembly let the swarm of robots manipulate spatially distributed objects. Clustering and assembling of objects is essential for construction processes.


These behaviors allow the coordinated movement of a swarm of robots in the environment.

  • Collective exploration navigates the swarm of robots cooperatively through the environment in order to explore it. It can be used to get a situational overview, search for objects, monitor the environment, or establish a communication network.
  • Coordinated motion moves the swarm of robots in a formation. The formation can have a well-defined shape, e.g., a line, or be arbitrary as in flocking.
  • Collective transport by the swarm of robots enables to collectively move objects which are too heavy or too large for individual robots.
  • Collective localization allows the robots in the swarm to find their position and orientation relative to each other via the establishment of a local coordinate system throughout the swarm.

Decision Making

These behaviors allow the robots in a swarm to take a common choice on a given issue.

  • Consensus allows the individual robots in the swarm to agree on or converge toward a single common choice from several alternatives.
  • Task allocation assigns arising tasks dynamically to the individual robots of the swarm. Its goal is to maximize the performance of the entire swarm system. If the robots have heterogeneous capabilities, the tasks can be distributed accordingly to further increase the system’s performance.
  • Collective fault detection within the swarm of robots determines deficiencies of individual robots. It allows determining robots that deviate from the desired behavior of the swarm, e.g., due to hardware failures.
  • Collective perception combines the data locally sensed by the robots in the swarm into a big picture. It allows the swarm to make collective decisions in an informed way, e.g., to classify objects reliably, allocate an appropriate fraction of robots to a specific task, or to determine the optimal solution to a global problem.
  • Synchronization aligns the frequency and phase of oscillators of the robots in the swarm. Thereby, the robots have a common understanding of time which allows them to perform actions synchronously.
  • Group size regulation allows the robots in the swarm to form groups of the desired size. If the size of the swarm exceeds the desired group size, it splits into multiple groups.

General Design Principles for Swarm Robots in Complex and Real-world Applications

To design and manage a wide range of possible systems, the key challenge will be to define a rigorous engineering methodology to program the individual robots so that the swarm as a whole acts as desired. This is no easy task because the characteristics of swarms at different scales might require radically different approaches.

  • Machine learning with data-driven approaches

The increasing complexity of swarm systems is such that their design cannot be accomplished solely by traditional approaches. The more robot swarms will be confronted with uncertain/unpredictable environments and will rely on intricate (i.e. that has many small parts or details) patterns of interactions, the more automated design methodologies will be necessary to obtain desired behaviors because they can be employed to generate individual rules that are evaluated for their effects on swarm performance. Machine learning with data-driven approaches becomes relevant whenever model-based solutions are too demanding, for instance, when it is difficult to provide a precise model of robot-environment interactions (e.g., due to complex physical interactions among micro/nanorobots or unpredictable environmental dynamics as caused by underwater currents). To address self-organized control, the learned control architectures need to suitably integrate robot perceptions with information asynchronously received from (possibly hundreds of) peers and with memory of past states. Hybrid systems mixing model-free and model-based approaches will likely provide additional power, learning from data those aspects that are peculiar to the problem at hand (e.g., the response of a noisy communication medium characterizing the target application). Overall, automated methods hold the potential to liberate the designer from tedious trial and error or parameter tuning and to better deal with the specific swarm scales required by the task, resulting in a general engineering methodology that can be suitably applied across different application domains.

  • Physical heterogeneity

In order for robot swarms to perform ever more complex tasks, they will likely need to be heterogeneous, both in hardware features and in roles that individual robots can play within the swarm. Physical heterogeneity can equip the robot swarm for tasks that require different hardware (e.g., fast-moving drones for collaborative monitoring that cooperate with slow ground robots capable of modifying the environment). Behavioral heterogeneity can enable specialization (possibly via local learning mechanisms) and can underlie phase transitions that make the emergent collective behavior more flexible, adaptive to new conditions, and resilient to external perturbations. By abandoning the traditional homogeneity of hardware and control, flexibility and autonomy can be improved for small and large swarms, and operations can be expanded to address a wider spectrum of spatial and temporal scales. The additional dimensions introduced by heterogeneity (i.e., number of different roles and their proportion within the swarm) entail increased complexity in the design and further motivate the exploitation of automated techniques.

  • Hierarchical forms of control

Robot swarms should include mechanisms to allow hierarchical forms of control beyond traditional pure self-organization. Whereas the latter features a flat organizational structure, a hierarchical control approach assigns a few individuals larger responsibilities, flexibly adapting the hierarchy to the task execution demands. In this way, it could be possible to more efficiently address task allocation, creation of task-oriented teams within the swarm, coordination of specific activities, or interaction with human users. Such hierarchies should, however, not be imposed from the outside. Rather, they should themselves be the result of self-organizing processes where some robots in the swarm may take leading roles depending on their specific characteristics, on the task being performed, and on the environmental conditions in which they are placed. In this respect, leveraging heterogeneity for robot swarm control — as advocated above — becomes even more consequential.

State of the Art from Research and Industrial Domains

Swarm Robots in Outer Space For Space Exploration

Since the beginning of space exploration, Mars and the Moon have been examined via orbiters, landers, and rovers. More than 40 missions have targeted Mars, and over 100 have been sent to the Moon. Space agencies continue to focus on developing novel strategies and technologies for probing celestial bodies. Multi-robot systems are particularly promising for planetary exploration, as they are more robust to individual failure and have the potential to examine larger areas.

Extraterrestrial exploration missions are increasingly directed toward challenging landscapes and environments, such as mountains, craters, lava tubes, and oceans. Rovers have been the most common vehicle choice for planetary exploration; however, they are designed to operate on relatively flat land. As a result, other types of robots, such as UAVs and hydrobots, are being proposed for more challenging environments and geographic features, including the atmospheres and oceans of celestial bodies. UAVs offer advantages over rovers for exploration, especially, where atmospheres are present: they provide higher resolution data than orbiters, have greater range and mobility, and can sample gases at different altitudes, thus also filling a planetary measurement gap.

Multirobot teams, perhaps with mixed capacities (that is, activators, sensors, and communication devices) could also be used to explore larger areas more effectively than single robots and could characterize and identify potential landing sites for manned missions as well as reveal hazardous areas. With an appropriate interface, a robotic team could conduct autonomous reconnaissance and increase human situational awareness of mission-critical information.

Swarm technologies, whereby systems of spacecraft or rovers (of varying degrees of intelligence) mimic swarms, colonies, or flocks in nature (such as bees, ants, or geese) have become a major focus for future NASA missions. These types of missions provide greater flexibility and the chance to gather more science than traditional single-vehicle missions. The emergent properties of swarms make these missions powerful, but at the same time more difficult to design and to ensure that the proper behaviors will emerge. An example of one of these missions is the NASA Autonomous NanoTechnology Swarm (ANTS) concept mission that will exploit swarm technology, launching 1000 picosatellites to investigate the asteroid belt, where a traditional spacecraft could not possibly survive. To examine an asteroid, the spacecraft will have to cooperate since they each have only a single instrument on board. To do this they will use an insect analogy of hierarchical social behavior where some spacecraft are directing others. Sub-swarms will exist that will act as teams that explore a particular asteroid based on the asteroid’s properties. Teams will have to share resources (instruments) with each other to investigate an asteroid.

The mission will also need to have collective or shared intelligence. Collective intelligence will also be needed so the same asteroids are not searched by spacecraft with the same instruments. This information may not be centralized but be in the form of the swarm’s collective knowledge. Collective intelligence will also need to be used when a team performs a virtual experiment. Spacecraft that map asteroids will have to share that knowledge with other spacecraft so they can perform their calculations on how to set up orbits or flybys. In addition, spacecraft will have to sequence themselves so that spacecraft do not run into each other or interfere with each other when they are performing their experiments.

Below is a brief survey of recent academic and industrial activities related to swarm intelligence robotics for space exploration:

  • Polytechnique Montreal Campus MIST Laboratory Swarm Robots

Professor Beltrame’s team at Polytechnique Montreal campus MIST Laboratory is carrying out a series of experiments with exploration-focused robots in order to prepare for three training missions planned for the coming months – a prelude to future exploration missions to the Moon and Mars.

They’re the size of a microwave oven but designed to do a job comparable to that of NASA’s large space rovers that have gone to Mars in recent years. Yet the main feature of robots is that they work in a team. In addition to covering a large area in a shorter period of time than a single rover would, being in a team enables each mission to better manage risk.

On the Moon or on Mars, there are no robot repair shops, so it’s important to have safety nets. If a robot breaks, the others can intervene to compensate for its loss.

The Polytechnique Montréal team has robot rovers in a variety of forms. In addition to the now-classic “wheeled” rovers, the research team is testing the exploration skills of a robot dog-style model. Also being assessed is the potential contribution of spherical robots that bounce as they move. The latter is a project developed by an École de technologie supérieure (ÉTS) research team headed by Professor David St-Onge.

  • Spacebit’s Walking Rover

UK-based Spacebit announced its plans to launch the United Kingdom’s first privately built lunar rover in 2021. The company entered a new partnership with the International Astronautical Federation Regional Group for Latin America and the Caribbean

Spacebit is offering robots with legs, which would allow the machines to delve into cracks and crevices inaccessible to traditional space rovers. The mission design calls for a rover to bring as many as eight such robots to a drop-off point. Then they would leave the mothership and in a swarm, explore lunar caves using artificial intelligence to bring back more details about the Moon’s history.

The London-based company intends to launch the first batch of these rovers together with Astrobotic’s Peregrine lunar lander. It will be the first lunar robotic explorer to rely on four legs rather than wheels to get around. These legs will allow the rover to explore lunar lava tubes, something that has never before been possible. With a sensor and two cameras, the rover will be able to gather exploration data and measurements on these tubes and other features on the lunar surface.

The Walking Rover’s design calls for a lightweight vehicle that weighs between 1 and 1.3 kg (2.2—2.85 lbs), relies on a combination of solar and battery power, and is operated by swarm intelligence. It will also be built to withstand the massive temperature variations that regularly occur on the lunar surface — 140 °C down to −171 °C.

  • Swarm Navigation and Exploration for Planetary Surface Missions

Swarm navigation is understood as a computationally data-driven technique to optimize swarm movements: either for location-aware formation control to optimize self-localization, or to span a distributed phased array to determine the bearing of a low-frequency radio signal, e.g. to navigate back to the lander.

In a study on swarm navigation and exploration for planetary surface missions, researchers propose a concept for joint wireless communication, localization, and timing, enabling robot self-localization and time synchronization. The multi-user radio channel access is achieved with a decentralized time division multiple access (TDMA) scheme: no central coordinator is required. The design of the wireless system is based on orthogonal frequency division multiplex (OFDM) to combat multipath and enable high-rate communication among agents. The wireless signals used for communication are further exploited to estimate the distance among agents based on the round-trip time. Ad-hoc communication and precise ranging provide the basis for distributed localization of all agents in a Bayesian tracking framework realized as distributed particle filters. As a result, the formation of the swarm is obtained. In addition, they exploit so-called multi-mode antennas (MMA; a multiport antenna, where different characteristic modes are excited independently) for 3D bearing determination. As a result, once agents in a swarm establish a network and are able to localize themselves, they can exploit this information to navigate and explore.

  • Lunar Zebro

The Lunar Zebro is an object-centered research project. A student team of 50 to 60 engineers and researchers with different backgrounds are working together closely to develop the smallest and lightest rover that can explore the surface of the Moon autonomously. The maximum mass of 1,5 kg and its survivor skills in harsh environments are the most challenging constraints the team has to work with. A Zebro is a six-legged swarm robot. The swarm can perform all sorts of complex search and reconnaissance tasks such as exploration of inhabitable terrains, tunnels, and caves or form a self-deploying and self-repairing sensor network. The idea is that a space mission becomes more effective if an Internet of (Swarm) Robots support bigger rovers and astronauts.
Many-fold innovations are being worked on in order for the Zebro to be Lunar-ready. Challenges the teams are solving are: how to program the Zebro in such a way that it will not harm itself or others when the software is affected by radiation; how the Zebro’s can operate autonomously within a specific area on a specific task and how they communicate this amongst themselves; how the structure of the Zebro can withstand the razor-sharp lunar dust; how the Zebro can interact with humans on a basis of remote command (not control); the flexibility of the sensors or extra batteries that a Zebro can carry and change when necessary.

The power of the concept manifests itself through the collective. Individually, the rovers have a simple design and are highly customizable. But collectively, they can in theory accomplish complex tasks. Zebros can work in a swarm, each robot making autonomous spur-of-the-moment (i.e. done without planning in advance) decisions while the collective is achieving a common goal. And because they are cheap, losing one robot is not the end of the mission.

  • Locust-inspired Jumping Moon Robot Swarm

The main implementation has relied on more traditional automobile-based rovers; however, this method is significantly limited by maneuverability. Unfortunately, rough terrain, obstacles, dust storms, and environmental conditions make it challenging for traditional land rovers to efficiently explore and cover large amounts of terrain. Applying these land rovers has opened many exploration and data collection opportunities, but the rate at which these have been carried out leaves more to be desired, especially due to a large number of unreachable areas. Considering this, the implementation of a novel bioinspired design focused on the jumping locomotion of locusts (i.e. a group of certain species of short-horned grasshoppers) may be advantageous as such a design could reduce the limitations of exploration by vastly expanding the movement capabilities of rover systems on the Moon. In conjunction with this expanded range of movement, this design would be compact and lightweight allowing for the transportation of several units designed to operate in a swarming fashion. These swarm mechanics would allow for communication between members to more effectively perform missions over a large area, such as terrain mapping, or lava tube exploration.

The swarm of the proposed bioinspired concept:

  1. has significant merit and enables dispersed (i.e. distribute or spread over a wide area) measurements at different cave and lava tube locations on the Moon;
  2. has the capability to obtain thermal information, temperature, radiation levels, and geothermal heat sources;
  3. can obtain images of walls, ceiling structures, and floors in the caves and lava tubes;
  4. is able to obtain information and images of any water sources and subterranean aquifers (i.e. an underground layer of water-bearing permeable rock, rock fractures or unconsolidated materials (gravel, sand, or silt)).

To ensure the success of NASA’s Artemis mission, the Artemis team must have enough information to complete the mission objectives. Before and after manned missions to the Moon, information about the lunar surface is vital to scout proper landing, architectural, and geological sites. The mechanics and size of the Lunar Locust robot would give it the ability to explore areas with rough terrain and where traditional locomotion is ineffective. To complete this goal, the NASA’s Lunar Reconnaissance Orbiter satellite is planned to provide surface data on the lunar environment where the Artemis III mission is set to land. In conjunction with this orbital satellite, the Lunar Locusts will be beneficial in characterizing and documenting the lunar geology on the Moon’s south pole.

  • Swarmathon: A Swarm Robotics Experiment for Future Space Exploration

In an effort to encourage the study and development of this technology, in 2016, the Swarmathon competition project was created from a cooperative agreement between the NASA Minority University Research and Education Program (MUREP) and the University of New Mexico (UNM) in partnership with the NASA Kennedy Space Centre Swamp Works. This annual competition is designed to engage students in developing cooperative robotic algorithms to revolutionize space exploration.

Since 2016 for four years, students from the University of Houston Clear Lake (UHCL) and San Jacinto College (SJC) have teamed up to compete in the yearly NASA Swarmathon contest. Three robots were provided for each Swarmathon team by the University of New Mexico and NASA. A robot is comprised of a mobile base mounted with navigation and guidance sensors. The body of the Swarmie is built from 3D printed and laser-cut parts designed for attaching and anchoring the hardware.

A small band of NASA engineers and interns is about to begin testing a group of robots and related software that will show whether it’s possible for autonomous machines to scurry about an alien world such as the Moon searching for and gathering resources just as an ant colony does.

Building on the research conducted at the University of New Mexico, the engineers at NASA’s Kennedy Space Center in Florida have been developing programs that tell small, wheeled robots to go out in different directions and randomly search an area for a particular material.

For these tests that are meant only to prove the software works and the concept is worthwhile, the robots are not searching for anything more than barcoded pieces of paper. In the future though, robots working around an asteroid or on the Moon or Mars would be equipped to scan the soil for infinitely valuable water-ice or other resources that can be turned into rocket fuel or breathable air for astronauts.

For now, the testing is limited to parking lots around Kennedy’s Launch Control Center using four homemade robots called “swarmies” that resemble stripped-down radio-controlled trucks. There are four of them, each with a webcam, WiFi antenna, and GPS device. They are being programmed to work on their own to survey an area, then call the others over when they find a cache of something valuable. It’s identical to the way an ant colony gathers around a food source to divide up the task of collecting the food and taking it back to the nest.

  • OffWorld: Swarm Robots for Ice Extraction in Space

California-based startup OffWorld has plans to make resource mining a reality across the Solar System. Its plan is to send swarms of smart robots to the surface of distant moons and planets to extract resources including water, in the form of ice, and minerals. The first thing OffWorld plans to do on the Moon is extract water ice for applications ranging from producing drinking water for humans to making rocket fuel. They operate in swarms, collaborating together, making decisions on their own, they can sense where the minerals and ore exist and act accordingly. Powered by solar electricity, these robots would be able to learn on the fly with little human intervention. They would also include modular construction so that similar parts could be used for different kinds of robots — whether in space or on the ground. Eventually, these robots could even self-replicate by building other robots using the local resources available on the Moon, Mars, or other destinations.

OffWorld has undertaken extensive Research and Development in the field of extreme environment (i.e. a habitat that is considered very hard to survive) industrial robotics initially applied to the mining and processing sector. The objective is the establishment of an end-to-end collaborative robotic system comprising hundreds or even thousands of multi-species robots working together with internal and collective autonomy to achieve defined strategic objectives. With the ongoing input of mining industry expertise on a daily basis, OffWorld has developed its robotic systems hand in hand with leading-edge know-how from the mining sector.

Key to the future of operations in space is the ability for robotic systems to undertake multiple complex tasks autonomously and with minimal human intervention. OffWorld has been developing a task agnostic machine learning framework to address and optimize any industrial process. This novel approach to minimally supervised autonomy ushers in a new era of remote operations in extreme environments such as the Lunar or Martian surface.

  • Marsbee: Swarm of Flapping Wing Flyers for Enhanced Mars Exploration

Another innovative project was accepted by the NASA Innovative Advanced Concepts (NIAC) program. Marsbee is a NASA-funded project which aims to send a swarm of robotic bees on Mars. The goal of these marsbees is to explore Mars, that is, for humans very difficult to do themselves. The exploration of the Red planet began a long time ago, but the area already investigated is not a large section. Hence, NASA decided to send robotic bees to explore the planet. These robotic bees known as, marsbees, are a size of a bumblebee with flapping wings attached to them. The size of their wings is larger than that of bumblebees to compensate for the thinner atmosphere. To explore the area efficiently, sensors are integrated into each marsbee. Through these sensors, they can investigate their environment, for example, measuring the temperature or humidity, look for types of obstacles, explore the type of surface, search for water and food resources, and so on. A mobile base station is used to launch and recharge the marsbees. It is also used as a communication interface between marsbees and the main base. Marsbees communicate wireless between themselves and with the base station.

The proposed architecture consists of a Mars rover that serves as a mobile base and a swarm of Marsbees. The swarm of Marsbee can significantly enhance the Mars exploration mission with the following benefits:

  1. Facilitating reconfigurable sensor networks;
  2. Creation of resilient systems that protects its critical capabilities (and associated assets) from harm;
  3. Sample or data collection using single or collaborative Marsbees.

Key technical innovation includes the use of insect-like compliant wings to enhance aerodynamics and a low power design. High lift coefficients will be achieved by properly achieving dynamic similarity between the bioinspired insect flight regime and the Mars environment.

From a systems engineering perspective, the Marsbee offers many benefits over traditional aerospace systems. The smaller volume, designed for the interplanetary spacecraft payload configuration, provides much more flexibility. Also, the Marsbee inherently offers more robustness to individual system failures. Because of its relatively small size and the small volume of airspace needed to test the system, it can be validated in a variety of accessible testing facilities.

Concluding Remarks

Robotic technology such as swarm robotics missions, automatic probes, and unmanned observatories allow for space exploration without risking human lives. Swarm technologies hold promise for exploration and scientific missions that require capabilities unavailable to missions designed around single spacecraft. Although swarm autonomy is clearly essential for missions where human control isn’t feasible, individual autonomy is essential for the survival of individual spacecraft as well as the entire swarm in hostile space environments. The derivative benefits of swarm computing require advances in miniaturization and nanotechnology. Moreover, the need for more efficient onboard power generation and storage motivates research in solar energy and battery technology, and the need for energy-efficient propulsion motivates research on solar sails and other technologies, such as electric-field propulsion. Groundbreaking advances in swarm intelligence research at NASA and other organizations (both civilian and military) show great possibilities for applying swarm computing beyond just space exploration.

Multi-spacecraft and swarm missions could change the risk-posture of future space exploration missions by affording loss of one or more agents without compromising the whole missions and also allow concurrent measurements and scientific explorations that are not possible using a monolithic architecture, e.g. interferometry (i.e. a measurement method using the phenomenon of interference of waves, usually light, radio or sound waves) using formation flying spacecraft, scientific exploration of comets and asteroids using a swarm fly-by, or exploration of the Moon and Mars lava tubes using a swarm of small rovers. In order to achieve technological readiness for such missions, we need to address the technology gaps such as resource-aware and network-aware autonomous task identification and task allocation for robot teams. Algorithms for optimizing what and when to communicate among assets, given the costs of the communication and the benefits of coordination. Relative localization/team member pose estimation from on-board sensors and subject to computational and network conditions of small spacecraft. Mission planning and scheduling that accounts for multiple dynamic assets; synchronization and/or distribution of plans. Human interfaces and autonomy software designed for an updated operations paradigm; overall, great individual autonomy will be needed as human sequencing for all agents is likely too cumbersome/impractical. Smaller and cheaper communications and sensor equipment, shifting the focus from individual robustness to redundancy.