What is Swarm Robotics?
A protocol about workshop 3 of the Hegau Bodensee Seminar,
presented by Amy Csorba-Prolić
Hegau Bodensee Seminar
in cooperation with the Zukunftskolleg at the University of Konstanz
University Day 2024 “Exploring Nature Around the World”
Friday, 15 November 2024, 9am - 1pm
Attendees: Swadhin Agrawal (ZUKOnnect Fellow from India/ Informatics)
and pupils from different schools of Konstanz and Singen
What is the Zukunftskolleg?
The Zukunftskolleg is an Institute for Advanced Study (IAS) at the University of Konstanz. It offers 2-year and 5-year fellowships for early career researchers (Postdocs). The Zukunftskolleg’s 5i strategy fosters early independence with an international, intergenerational, intra-university and interdisciplinary angle. With its ZUKOnnect Fellowships (Zukunftskolleg Konnect Fellowships), the Zukunftskolleg supports early career researchers from Africa, Asia and Latin America, who do research that fits to one of the thirteen departments at the University of Konstanz.
“What is Swarm Robotics”?
Firstly Swadhin Agrawal, who is a ZUKOnnect Fellow and host of this workshop, briefly talked about being a researcher in swarm robotics. His current research focuses on Swarm intelligence and Human Interactions. He has finished his Bachelors and Masters in Physics (Major) and Electrical Engineering and Computer Sciences (Minor) in the year 2020 from IISER Bhopal. In addition he also told us that he was from India from the City of Lakes Bhopal where he also went to university.
Secondly, he showed us what caused the invention of robots and that they were actually a fictional idea that had been used in fictional movies in the 1920s. Then those fictional ideas turned into a real working machine, the Turing machine by Alan Turing in 1936. Though they stopped inventing robots in the 1970s until the 21st century (AI winter) because humans were terrified about their loss of control and that robots eventually would no longer be able to be controlled by humans. This continued until people started to research and discovered Biological Neurons. That was when they finally started to get back to inventing and discovering new things about robots and Artificial Intelligence. Now research and new technology is being improved continuously.
More importantly we defined the term robot which we did by him asking us what we think a robot is and afterwards we clarified this definition. He told us that the word robot came from the Czech word robota which means forced or slave labourer and that in 1921, Karl Capek used this term as the first one and he defined a robot by saying that a robot is a machine that looks similar to a human. Even though the real definition of a robot is a machine that can move independently (act) and perform complex actions similar to a living creature, a robot has to be able to plan, sense and act. Those functions are the key to a robot who is able to do any given task on its own because of its abilities.
A robot has to have the ability to move independently which requires the robot to act. It has to have the ability to perform complex tasks, so the robot does more than a simple repetitive motion (it must have some ability to think- to determine the best sequence of actions to perform) in order to complete a task by making decisions, which means they have to plan. Thirdly, they have to have the ability to sense their surroundings and their own position, which is the first and most important part of its ability by gathering different kinds of information.
This also is used in coding for robots by breaking a task down. For example in coding there are four parts that have to exist in order for the robot to get instructions. Firstly, it has to determine where it is which is done by giving the robot information like a map and its position. Then it has to know where it has to go. This step requires the robot to have a goal but also it has to send commands to his motors to make the robot move. Afterwards it has to know where to go next and lastly it has to know when the task is complete. As we learned before, the importance of a robot is its ability to sense itself and its surroundings and there are two ways it is able to sense. The first one is sensing actively so by sending signals to its environment and then through sensors it is able to get feedback.
But there is another way of sensing which is passive sensing where the robot only gets feedback from its surroundings. After that we’ve learned about what a basic control loop is and how it works. So a basic control loop is designed to maintain the stability of a system and to produce a consistent outcome that is desired. A control loop begins with input of the robot secondly this input goes to the system of the robot. Then it takes the wanted action and if there is any missing information or errors it will try to get more input and a loop is created until no more errors show up but usually it is a never ending loop or until the task is complete. Another way of control loop is a loop where there is an input, then a system, afterwards a computer and then the action. Those types of loops are called algorithms like AI (Artificial intelligence).
Also there is a dampening that is required for a robot to achieve its task or rather to meet its setpoint. This damped is named critically damped however two other dampers that we don't want are over or under damped. Further he told us that there are multi-robot systems which have different robots that all do different parts of a task but overall are working together but those kinds of systems have single point failure which means that when any of those robots isn’t able to finish its part of the task the whole task fails so they are more likely to fail a task. Though there is a different system called robot swarms where all of the robots have the same mission which technically means they all have the same task but there are difficulties too. For example they are difficult to design and to control because when you get those swarms you are only able to control one of them, which is a huge problem because they are all individuals. And mostly because we don't know how they would react or what they would do in this type of system due to its randomness, people like Swadhin are researching exactly that. But what we could do with swarm robotics is incredible. We would be able to cure illnesses where we as humans aren't able to operate, to be able to clean the ocean in an efficient way and to explore the deep levels of the ocean where humans aren't able to go to.
Afterwards we did a fun game which showed us how swarms work. This game began by standing in a tight circle where we had to put both of our hands in the middle, then we had to randomly grab two hands and wait until all of us had both hands locked with two others. Now we had the task to untangle each one to make a big untangle circle and firstly with speaking and in the second round without speaking. In the first round we sadly untangled ourself but managed to make two circles instead of one which was’t our mission but in the second round we managed it and we all were all really happy.
Then we looked at collective motion models. The first one was a topological interaction model which is based on a leader and their focus lies in staying connected even though long distances. And secondly we looked at metric-based zonal models based on their surroundings where they have zones to communicate, so they are dependent on the physical distance of each other. Those two models show how different swarms can behave inorder for them to reach a certain goal or finish a task.
A similar type of model is the defection in animal groups where we have cooperators (disoriented followers) and defectors (leaders). Those leaders know where to go and minimize the damping and by aligning them, defecting makes reaching a goal/ task a lot faster. But we have also learned that there are different types of defection, though first we must understand the cousin model which explains that there are three zones where cooperators can be in. They can be in zone a, o or r Zone a means that they won't return if they aren’t in the right direction because they are still near the defector. In zone o or r they will return because of their long distance to the defector.
Afterwards we went to the labour of Swadhin and saw the cute Swarmies, his robot swarm simulating robots. There we also were able to ask questions about how the swarm robotics work and how they are related to his already given presentation, though this was the end of this uni day.