On the Verge of a Breakthrough
The vision of robot swarms is becoming (albeit slowly) a reality: most of the future systems in the air, at sea and on the ground, as well as in space and cyberspace, will be unmanned, robotic, autonomous and intelligent. Exclusive
The PD-100 "Black Hornet" is a Nano-UAV that demonstrated stellar performance in a trial conducted in early 2015 at the US Army base in Fort Benning, Georgia. This UAV is the size of a man's finger. It weighs 18 grams and carries a day/night camera and a miniature transmitter that sends still and video images to the operator via a datalink. The PD-100 Black Hornet was designed to collect intelligence and can operate autonomously or semi-autonomously.
What is the major advantage of this miniature Black Hornet? It is a part of a robot system that consists of two UAVs and a control system. Everything is packaged in a backpack carried by a single trooper, and the over-all weight of the entire system is only 1.5 kg. The Black Hornet has already been used operationally by the 1st Intelligence, Surveillance & Reconnaissance Brigade of the British Army. Experts in robotics for military and security applications regard this system as the precursor of the next generation of robotic applications on the battlefield as well as in civilian life – no more single robots, but rather groups, flocks or swarms of robots.
Dr. Noah Agmon, researcher of robotics and lecturer at the Bar-Ilan University, disapproves of the term "swarm" as it reminds her of swarms of bees or ants. Admittedly, it seems that we have learned a lot from the behavior of animal swarms, but in the academic world, Dr. Agmon prefers to use such terms as a "Group" or a "Work Team" of robots.
What have we learned from nature? Studies have shown that when a bird sees five other birds ahead, it will decide to follow them. A school of fish in the ocean moves at a high speed as one when one of the fish decides to change direction in order to escape from a predator or reach a source of food. It is a known fact that in large flocks of birds that migrate to distances of hundreds or thousands of kilometers, there are several leaders that take the lead in rotation, as the first bird, the leader, sustains the strongest blast of air as it is positioned in front, so it rotates with several co-leaders. This is an example of decision making and communication between the individual members in a flock of birds.
In the human world and civilian life, we are already familiar with the robot groups of the Amazon Company that fetch book packages from their warehouses. In the military and security world we have seen the first steps in the employment of groups of robotic vehicles: in Afghanistan, the USA employed 12,000 robotic ground vehicles in search and rescue missions, for entering buildings, for searching suspected vehicles and for monitoring and bomb disposal. Those were not groups of autonomous vehicles that made their own decisions – Each vehicle was driven and commanded separately. In Israel, groups of robotic vehicles patrol the border along the fence surrounding the Gaza Strip, but these groups do not make their own decisions.
Dr. Noah Agmon explains: "As a researcher, I would have loved to see a greater degree of autonomy in the robotic vehicle groups, namely – the group must have a centralized system that makes decisions instead of a human element, as a human operator can control three to four vehicles and no more. Through our research, we hope to reach a situation where a single human operator in the loop will be able to command not just one vehicle but a hundred vehicles, with the vehicles employed to perform different and highly diversified tasks on the ground, in the air and at sea. The tools required for that purpose are easier and more simplified communication, smarter algorithms, more reliable software and more sophisticated sensors. All of that is required so that a group of 100 robots does not get stuck half way through the operation and can fully accomplish the mission assigned to it."
Researchers agree that the field of employing groups of autonomous robots in military and civilian applications has a huge potential. Some breakthroughs have been made, but in general, the field is still in its infancy. The future use of military applications is very promising. Researcher Shahar Sarid from the robotics laboratory at the Ben-Gurion University of the Negev claims that "Mobile robots perform important tasks in military and police applications and in military reconnaissance vehicles (surveillance), and there are robots used for bomb disposal. In the civilian field – (they are used) for search and rescue operations, for cleaning buildings and swimming pools and for mowing lawns."
Dr. Noah Agmon envisions groups of robots used for intelligence gathering, border patrol, and search and rescue missions; groups of robots dispatched to detect minefields or to scan suspected buildings before the troopers enter them. Other groups may be dispatched to chart urban territories.
Ofer Haruvi, Head of Divisional R&D at IAI's Military Aircraft Division envisions the future military application of robot groups as follows: "You will want, for example, to launch several sensor types to various targets. But you cannot load too many sensors onto a single UAV. So you will divide the sensors among several robotic vehicles. Some of those vehicles will be airborne while others will operate on the ground, or at sea, and they will cooperate – the airborne vehicles will look ahead while the ground vehicles or the sea vehicles will follow their predetermined routes. This is an imaginary example of how a group of robots may be utilized for military needs. In civilian life, like in agriculture, they think about employing multiple airborne robots, with one group dealing with pest sensing and another group spraying crops."
Professor Amir Shapira, head of the robotics laboratory at the Ben-Gurion University of the Negev, stresses that systems of 100 robots are already used for cooperative charting and navigation, and in his view "The activity of employing robot groups is making giant leaps forward and soon we can expect a major breakthrough by the entire field of robotics. Giant corporations such as Google, Amazon and Apple are keenly interested in this field. A lot of progress has been made with regard to all of the elements of the operational doctrine for robot groups, and the doctrine is based on numerous elements, notably software, algorithms, hardware, mechanics, electronics and communication."
In 2014, the Israel Institute for National Strategic Studies (INSS) published a forecast for unmanned systems, envisioning "Hundreds of airborne vehicles attacking targets in Lebanon, Syria and Iran, a 'blitz against the Axis of Evil'. Battalions of humanoid robots and swarms of robot-birds used for intelligence gathering, dozens of UAVs charged with solar energy employed for intelligence gathering and battle damage assessment missions."
The periodical IEEE Spectrum reported an exercise conducted on the James River, Virginia by the US Navy Office of Naval Research (ONR). The name of the exercise was Robot Boat Swarms, and the reporter described "a squadron of robotic vessels approaching the enemy vessels like a school of sharks around its prey."
15 autonomous vessels took part in the exercise. Their task was to defend the flagship. The exercise planners used a command and control architecture to operate the command and sensing systems of the autonomous vessels so that they may coordinate their operations. The source of the technology used was a software originally used by NASA on space flights. Each vessel transmitted the picture picked up by its radar to the other vessels, so the entire group of vessels shared the same situation awareness and they could navigate cooperatively. Following this exercise, it was decided that the technology known as Robotic Swarm Technology will be used by the US Navy in the future. Paul Scharre, a senior researcher at the Center for a New American Security in Washington had this to say in this context: "Since World War II we kept saying that it was not quantity that mattered but quality. However, in today's wars, quality is not enough, and in order to achieve overbalance, quantity will be required as well. Unmanned systems in large numbers can help achieve overbalance on the battlefield. We are talking about large groups of inexpensive platforms, and in the US Navy they are currently hard at work adapting advanced sensing systems for autonomous vessels, so that those vessels can 'see' their environment and operate in it. There is one reservation, however, it was determined that the decision as to whether to open fire on the enemy targets will still be made by humans."
The Group as a Single Entity
Students at the School of Engineering of Harvard University developed, about a year ago, a group of 1,000 miniature robotic vehicles, each a few centimeters long with three tiny legs. An order was issued to the group, and the robots started emitting light and lined up in a five-point star formation. The Harvard engineers summarized the experiment in a description published by the periodical Science as follows: "The beauty of the systems is in their simplicity. At some point you no longer see the individual robot but the entire group as a single entity."
The conclusion was that the new technologies for the employment of robot groups all stem from previous technologies of developments that already exist, as cruise missiles that navigate and steer themselves over enormous distances, bomb disposal robots, various types of UAVs and MUAVs. Everything is based on the assumption that artificial intelligence will be able to perform such cognitive tasks as learning from previous mistakes and coordinating the activities of multiple units.
Experts in the USA see significant advantages in the development of systems made up of groups of robots for air, sea and air-defense applications: unmanned platforms are inexpensive. A robot can operate much longer than a human, and it is not sensitive to rough living conditions in the air or at sea. As far as anti-aircraft and anti-missile defense is concerned, groups of robots defending an objective will be able to execute evasive maneuvers human pilots or even missiles cannot execute. With regard to attacks against enemy air-defense layouts – the operations of the robot groups can be planned so that the group may sacrifice a few robots so that all the other may gain access and execute the attack against the enemy air-defense system. No military in the world will even consider including such sacrifices in its plans for attacks by manned fighter aircraft.
An article published in Breaking Defense went even further by stating that the US military is pressing for the development of robot groups, but not just robots for "Dirty, Dull and Dangerous" missions like bomb disposal, but also for such missions where the robots will be called upon to actually destroy enemy targets. "The military wants mechanized mules that would carry loads and help the warfighters to fight. It wants small robots that would help the warfighter identify obstacles and threats along his axis of advance. It wants UAVs that would fly ahead of the attack helicopter and report what goes on at the front line to the pilots. The trend should be 'remove the humans and save on personnel and human lives'. The vision is that most of the future systems in the air, at sea and on the ground, as well as in space and in cyberspace, will be unmanned, robotic, autonomous and intelligent."
Regrettably, we are unable to reveal what the Israeli defense establishment and defense industries are doing in the field of robot groups for military/defense applications. All of the defense industries flatly refuse to even discuss the matter, which is highly classified. Nevertheless, Ofer Haruvi says that "Over the last decade there has been intensive activity in this field in the academic world, with regard to civilian and military aspects. There are numerous research projects under way involving robot groups and quadcopters, for example, that fly in groups and perform various tasks. We at the industry do not ignore the academic activity and will not lag behind it."
Haruvi says that it is clear how a single UAV operates and what it can do – but how do 40 UAVs, all airborne at the same time and flying together, operate? "How do we see to it that they do not collide with one another? We should develop such tools as algorithms and numerous simulations. The challenge is developing the operational concept or doctrine for a group of robots. How can a single person control 40 vehicles? Publications are currently available with regard to two possible scenarios: one – I need to cover a large territory of thousands of acres in an effort to collect information for agricultural purposes, for example. For that purpose, I will employ multiple robots, each one of which will perform the same activity, and at the end of the process I will integrate all of the information they collected. Another scenario: each robot will perform a different activity as part of the effort to collect information about the given territory.
Haruvi concludes the robot group issue: "The foundation is a platform with a payload. The employment of large vehicle groups offers an advantage – the individual platform may be a little 'dumber' and less sophisticated in its capabilities, as the quantity makes up for the lack of quality."
At the robotics laboratory of the Bar-Ilan University, Dr. Noah Agmon, her fellow researchers and the students conduct experiments in the form of search missions by a group of 10 robots. There are several robot types at this laboratory, including types capable of cooperating with quadcopters – air-to-ground cooperation by robots. In one room, two robots move across the floor. Their task is to search for cylinders hidden around the room. Each robot performs the search individually, but when they approach one another, they keep their distance, avoid a collision, and each one makes way for the other, demonstrating excellent robotic manners.
Dr. Agmon tells us that "The world of autonomous robot groups is fairly new. The research activity may have progressed further than the industrial implementation. One should bear in mind the fact that this field of research is only about 15 years old, but the civilian and military potential is tremendous. There is close, intensive cooperation between the academic world and the military, and the defense establishment has understood that it is mandatory to go in this direction. I have no doubt that this field will provide solutions, for example – for the charting of subterranean tunnels. If I dispatch a single robot into a tunnel it may get stuck or we may lose contact with it, but dispatching a group of robots into a tunnel would provide a much better chance of effectiveness. Today, everything is based on a single robot and a single operator, and that, naturally, is by no means the most efficient utilization of the robotic potential."
According to Dr. Agmon, the difficulties encountered in the development of robot groups involve the communication with them and among them, and the energy supply sources. "These things are difficult to perform under laboratory conditions, but we know that the military world has some solutions. A quadcopter can remain airborne for 20 minutes maximum – that is what its battery allows. Solutions are being sought, and I have no doubt that systems capable of operating over longer periods of time and with a high degree of reliability will be developed. In this particular field, we in Israel are second to none compared to the world."
The vision of robot swarms is becoming (albeit slowly) a reality: most of the future systems in the air, at sea and on the ground, as well as in space and cyberspace, will be unmanned, robotic, autonomous and intelligent. Exclusive
The PD-100 "Black Hornet" is a Nano-UAV that demonstrated stellar performance in a trial conducted in early 2015 at the US Army base in Fort Benning, Georgia. This UAV is the size of a man's finger. It weighs 18 grams and carries a day/night camera and a miniature transmitter that sends still and video images to the operator via a datalink. The PD-100 Black Hornet was designed to collect intelligence and can operate autonomously or semi-autonomously.
What is the major advantage of this miniature Black Hornet? It is a part of a robot system that consists of two UAVs and a control system. Everything is packaged in a backpack carried by a single trooper, and the over-all weight of the entire system is only 1.5 kg. The Black Hornet has already been used operationally by the 1st Intelligence, Surveillance & Reconnaissance Brigade of the British Army. Experts in robotics for military and security applications regard this system as the precursor of the next generation of robotic applications on the battlefield as well as in civilian life – no more single robots, but rather groups, flocks or swarms of robots.
Dr. Noah Agmon, researcher of robotics and lecturer at the Bar-Ilan University, disapproves of the term "swarm" as it reminds her of swarms of bees or ants. Admittedly, it seems that we have learned a lot from the behavior of animal swarms, but in the academic world, Dr. Agmon prefers to use such terms as a "Group" or a "Work Team" of robots.
What have we learned from nature? Studies have shown that when a bird sees five other birds ahead, it will decide to follow them. A school of fish in the ocean moves at a high speed as one when one of the fish decides to change direction in order to escape from a predator or reach a source of food. It is a known fact that in large flocks of birds that migrate to distances of hundreds or thousands of kilometers, there are several leaders that take the lead in rotation, as the first bird, the leader, sustains the strongest blast of air as it is positioned in front, so it rotates with several co-leaders. This is an example of decision making and communication between the individual members in a flock of birds.
In the human world and civilian life, we are already familiar with the robot groups of the Amazon Company that fetch book packages from their warehouses. In the military and security world we have seen the first steps in the employment of groups of robotic vehicles: in Afghanistan, the USA employed 12,000 robotic ground vehicles in search and rescue missions, for entering buildings, for searching suspected vehicles and for monitoring and bomb disposal. Those were not groups of autonomous vehicles that made their own decisions – Each vehicle was driven and commanded separately. In Israel, groups of robotic vehicles patrol the border along the fence surrounding the Gaza Strip, but these groups do not make their own decisions.
Dr. Noah Agmon explains: "As a researcher, I would have loved to see a greater degree of autonomy in the robotic vehicle groups, namely – the group must have a centralized system that makes decisions instead of a human element, as a human operator can control three to four vehicles and no more. Through our research, we hope to reach a situation where a single human operator in the loop will be able to command not just one vehicle but a hundred vehicles, with the vehicles employed to perform different and highly diversified tasks on the ground, in the air and at sea. The tools required for that purpose are easier and more simplified communication, smarter algorithms, more reliable software and more sophisticated sensors. All of that is required so that a group of 100 robots does not get stuck half way through the operation and can fully accomplish the mission assigned to it."
Researchers agree that the field of employing groups of autonomous robots in military and civilian applications has a huge potential. Some breakthroughs have been made, but in general, the field is still in its infancy. The future use of military applications is very promising. Researcher Shahar Sarid from the robotics laboratory at the Ben-Gurion University of the Negev claims that "Mobile robots perform important tasks in military and police applications and in military reconnaissance vehicles (surveillance), and there are robots used for bomb disposal. In the civilian field – (they are used) for search and rescue operations, for cleaning buildings and swimming pools and for mowing lawns."
Dr. Noah Agmon envisions groups of robots used for intelligence gathering, border patrol, and search and rescue missions; groups of robots dispatched to detect minefields or to scan suspected buildings before the troopers enter them. Other groups may be dispatched to chart urban territories.
Ofer Haruvi, Head of Divisional R&D at IAI's Military Aircraft Division envisions the future military application of robot groups as follows: "You will want, for example, to launch several sensor types to various targets. But you cannot load too many sensors onto a single UAV. So you will divide the sensors among several robotic vehicles. Some of those vehicles will be airborne while others will operate on the ground, or at sea, and they will cooperate – the airborne vehicles will look ahead while the ground vehicles or the sea vehicles will follow their predetermined routes. This is an imaginary example of how a group of robots may be utilized for military needs. In civilian life, like in agriculture, they think about employing multiple airborne robots, with one group dealing with pest sensing and another group spraying crops."
Professor Amir Shapira, head of the robotics laboratory at the Ben-Gurion University of the Negev, stresses that systems of 100 robots are already used for cooperative charting and navigation, and in his view "The activity of employing robot groups is making giant leaps forward and soon we can expect a major breakthrough by the entire field of robotics. Giant corporations such as Google, Amazon and Apple are keenly interested in this field. A lot of progress has been made with regard to all of the elements of the operational doctrine for robot groups, and the doctrine is based on numerous elements, notably software, algorithms, hardware, mechanics, electronics and communication."
In 2014, the Israel Institute for National Strategic Studies (INSS) published a forecast for unmanned systems, envisioning "Hundreds of airborne vehicles attacking targets in Lebanon, Syria and Iran, a 'blitz against the Axis of Evil'. Battalions of humanoid robots and swarms of robot-birds used for intelligence gathering, dozens of UAVs charged with solar energy employed for intelligence gathering and battle damage assessment missions."
The periodical IEEE Spectrum reported an exercise conducted on the James River, Virginia by the US Navy Office of Naval Research (ONR). The name of the exercise was Robot Boat Swarms, and the reporter described "a squadron of robotic vessels approaching the enemy vessels like a school of sharks around its prey."
15 autonomous vessels took part in the exercise. Their task was to defend the flagship. The exercise planners used a command and control architecture to operate the command and sensing systems of the autonomous vessels so that they may coordinate their operations. The source of the technology used was a software originally used by NASA on space flights. Each vessel transmitted the picture picked up by its radar to the other vessels, so the entire group of vessels shared the same situation awareness and they could navigate cooperatively. Following this exercise, it was decided that the technology known as Robotic Swarm Technology will be used by the US Navy in the future. Paul Scharre, a senior researcher at the Center for a New American Security in Washington had this to say in this context: "Since World War II we kept saying that it was not quantity that mattered but quality. However, in today's wars, quality is not enough, and in order to achieve overbalance, quantity will be required as well. Unmanned systems in large numbers can help achieve overbalance on the battlefield. We are talking about large groups of inexpensive platforms, and in the US Navy they are currently hard at work adapting advanced sensing systems for autonomous vessels, so that those vessels can 'see' their environment and operate in it. There is one reservation, however, it was determined that the decision as to whether to open fire on the enemy targets will still be made by humans."
The Group as a Single Entity
Students at the School of Engineering of Harvard University developed, about a year ago, a group of 1,000 miniature robotic vehicles, each a few centimeters long with three tiny legs. An order was issued to the group, and the robots started emitting light and lined up in a five-point star formation. The Harvard engineers summarized the experiment in a description published by the periodical Science as follows: "The beauty of the systems is in their simplicity. At some point you no longer see the individual robot but the entire group as a single entity."
The conclusion was that the new technologies for the employment of robot groups all stem from previous technologies of developments that already exist, as cruise missiles that navigate and steer themselves over enormous distances, bomb disposal robots, various types of UAVs and MUAVs. Everything is based on the assumption that artificial intelligence will be able to perform such cognitive tasks as learning from previous mistakes and coordinating the activities of multiple units.
Experts in the USA see significant advantages in the development of systems made up of groups of robots for air, sea and air-defense applications: unmanned platforms are inexpensive. A robot can operate much longer than a human, and it is not sensitive to rough living conditions in the air or at sea. As far as anti-aircraft and anti-missile defense is concerned, groups of robots defending an objective will be able to execute evasive maneuvers human pilots or even missiles cannot execute. With regard to attacks against enemy air-defense layouts – the operations of the robot groups can be planned so that the group may sacrifice a few robots so that all the other may gain access and execute the attack against the enemy air-defense system. No military in the world will even consider including such sacrifices in its plans for attacks by manned fighter aircraft.
An article published in Breaking Defense went even further by stating that the US military is pressing for the development of robot groups, but not just robots for "Dirty, Dull and Dangerous" missions like bomb disposal, but also for such missions where the robots will be called upon to actually destroy enemy targets. "The military wants mechanized mules that would carry loads and help the warfighters to fight. It wants small robots that would help the warfighter identify obstacles and threats along his axis of advance. It wants UAVs that would fly ahead of the attack helicopter and report what goes on at the front line to the pilots. The trend should be 'remove the humans and save on personnel and human lives'. The vision is that most of the future systems in the air, at sea and on the ground, as well as in space and in cyberspace, will be unmanned, robotic, autonomous and intelligent."
Regrettably, we are unable to reveal what the Israeli defense establishment and defense industries are doing in the field of robot groups for military/defense applications. All of the defense industries flatly refuse to even discuss the matter, which is highly classified. Nevertheless, Ofer Haruvi says that "Over the last decade there has been intensive activity in this field in the academic world, with regard to civilian and military aspects. There are numerous research projects under way involving robot groups and quadcopters, for example, that fly in groups and perform various tasks. We at the industry do not ignore the academic activity and will not lag behind it."
Haruvi says that it is clear how a single UAV operates and what it can do – but how do 40 UAVs, all airborne at the same time and flying together, operate? "How do we see to it that they do not collide with one another? We should develop such tools as algorithms and numerous simulations. The challenge is developing the operational concept or doctrine for a group of robots. How can a single person control 40 vehicles? Publications are currently available with regard to two possible scenarios: one – I need to cover a large territory of thousands of acres in an effort to collect information for agricultural purposes, for example. For that purpose, I will employ multiple robots, each one of which will perform the same activity, and at the end of the process I will integrate all of the information they collected. Another scenario: each robot will perform a different activity as part of the effort to collect information about the given territory.
Haruvi concludes the robot group issue: "The foundation is a platform with a payload. The employment of large vehicle groups offers an advantage – the individual platform may be a little 'dumber' and less sophisticated in its capabilities, as the quantity makes up for the lack of quality."
At the robotics laboratory of the Bar-Ilan University, Dr. Noah Agmon, her fellow researchers and the students conduct experiments in the form of search missions by a group of 10 robots. There are several robot types at this laboratory, including types capable of cooperating with quadcopters – air-to-ground cooperation by robots. In one room, two robots move across the floor. Their task is to search for cylinders hidden around the room. Each robot performs the search individually, but when they approach one another, they keep their distance, avoid a collision, and each one makes way for the other, demonstrating excellent robotic manners.
Dr. Agmon tells us that "The world of autonomous robot groups is fairly new. The research activity may have progressed further than the industrial implementation. One should bear in mind the fact that this field of research is only about 15 years old, but the civilian and military potential is tremendous. There is close, intensive cooperation between the academic world and the military, and the defense establishment has understood that it is mandatory to go in this direction. I have no doubt that this field will provide solutions, for example – for the charting of subterranean tunnels. If I dispatch a single robot into a tunnel it may get stuck or we may lose contact with it, but dispatching a group of robots into a tunnel would provide a much better chance of effectiveness. Today, everything is based on a single robot and a single operator, and that, naturally, is by no means the most efficient utilization of the robotic potential."
According to Dr. Agmon, the difficulties encountered in the development of robot groups involve the communication with them and among them, and the energy supply sources. "These things are difficult to perform under laboratory conditions, but we know that the military world has some solutions. A quadcopter can remain airborne for 20 minutes maximum – that is what its battery allows. Solutions are being sought, and I have no doubt that systems capable of operating over longer periods of time and with a high degree of reliability will be developed. In this particular field, we in Israel are second to none compared to the world."
