Can humans perceive autonomous balloons as intentional beings
Many intelligent robots have multiple functions to help people do much burdensome work or accompany people, which clearly indicates the increasing importance of the robot. In this paper, we explore the balloon movement as a form of intentional features communicating with people and set up the environment to conceive people to tell a story about it. We are curious as to how people might envision interacting with autonomous balloon constructing with simple rules in the closed space. According, we focused our attention on one autonomous balloon with four motors and propeller having ability to roam in the space freely, which allowed us to investigate the interaction between the movement of balloon and people. In this process, the two exhibition we set up support us to get twelve participant feedback about the thought of balloon’s movement. The qualitative findings that we present could let other people imagine the interaction between human and primitive form of autonomous robot.
Autonomous Balloon, Flocking algorithm, Perception, Emergent behavior.
How people feel like objects themselves that are alive, in other words, which have intentional behavior? Not through the perspective of intelligent objects, like robots, but through a perspective of daily life objects we see every day. Therefore, the simple rules i adopt that could create complex behavior which conceive people these agents are livable. Boid, a rule developed by Reynold who is invented the flocking algorithm, explores that a balloon moves according to a simple flocking rule that is constructed by three factors which are separation, alignment and cohesion. But from a simple object – a balloon – moves according to simple rules having similar characters with boids. According to studies that performed by Reynold, by using these simple rule movement of animated flocks, an object can be immediately recognized as a representation of a natural flock. The study also adds that these behaviours would be delightful to watch (Reynold, 1986). The project, QUANTUMminus is an interactive space inhabited by a flying balloon robot who playfully engages humans in a performative exchange. This project addresses the question of how a simple movement rule, like boid, can be perceived as an intentional movement.
The balloon’s behavior and personality aims to tempt us to collaborate and to form a shared space of experience(fig.1). By only mixing three simple flocking behavior rules, different impression on people can be achieved. The flocking algorithm of the balloon in the project QUANTUMminus is based on three simple rules which are collision avoidance, velocity matching and flocking centering. According to the ego tunnel of Thomas Metzinger(2017) that our sense of self is a virtual simulation within a larger simulation of an external reality which is larger still(Metzinger,2017). However, each boid has much limited sense of the world. “What we see and hear, or what we feel and smell and taste, is only a small fraction of what actually exists out there” (Metzinger, 2017). Regardless of limited tracking technology of Optitrack and the physical limitation of the helium balloon, we could adopt three basic rules from flocking algorithm rather than create the whole flocking phenomenon.
Figure.1: QUANTUMinus, the flocking balloon exhibited at Tate Bratain exhibition in July 2019.
In the project QUANTUMminus, balloon robots could move in 3D dimension space freely. Usually the autonomous robot system has certain degree of self-sufficiency. It has certain degree of capability that could make its own choice based on the sense of autonomous robot toward the environment. From creating autonomous vehicle systems, these robots take responsibility for themselves, which means that each agent has its own controller rather than controlled by leaders or global plans. For us, the autonomous robot did not just move. But, its important that the conceptual leap for acting in certain way has the motivation and ability depend on the environment. Therefore, an autonomous agent has three key characteristic from Autonomous Minimalist. Firstly, the agents need to perceive the surrounding with limited perception capability and then calculate all the elements in the environment next. Further on, the agents choose the action that comes from the calculation without a leader. Technically speaking, our flying balloon is an autonomous robot. The balloon which has no sensor attached does not have the ability to perceive the environment, it receives the command from computer (central controller). From Autonomous Minimalist Following In Three Dimensions: A study shows that by using small-scale dirigibles, we get to know the limitation of capability of helium balloon in order to decrease the weight. Therefore, we are using the exterior system to construct our autonomous balloon.(fig.2.1) Optitrack is used to localize balloons due to technical convenience. Therefore, the outcome simulation has similar performance as autonomous robot.
The flying balloon is animated based on the three flocking autonomous algorithmic steering behaviors. These behaviors allowed individual balloon to take action in their environments in a series of “lifelike” manners: fleeing, wandering, following, avoiding, thinking, floating. The strategies change based on the position of people in an environment.
Figure.2.1: The connection between the autonomous system and flying balloon exterior system.
In order to compare the behavior of balloons and boids, firstly we need to be familiar with boids’ behavior. The goal of this project is to allow users to sense that the balloons contain intentional behavior, autonomous motivation and unique personality. Among the agents, each boid is perceiving and making changes to the environment and also affecting the other agents. The first rule of flocking is collision avoidance, it avoids collisions from flockmates nearby; the second rule, flock centering attempt to match the velocity of nearby flockmates; the third rule, steer toward the average heading of neighbor.(fig.2.2)
“Perhaps most puzzling is the strong impression of intentional, centralised control. Yet all the evidence indicates that flock motion must be merely the aggregate result of the actions of individual animals, each acting solely on the basis of its own local perception of the world”(Reynolds, 1987). Then the useful and interesting behavior could be created through one agent sensing other agents, let alone the people also would be considered relative to these agents. In this section, the most important thing is our autonomous balloon only adapt three simple rules from flocking algorithm in choosing movement.
QUANTUMminus was exhibited the first time at Here East in project faire. The display of the project only allowed two persons to interact with one balloon. During this time QUANTUMminus was only developed using the one rule which is to stay close to the person, which is called perceived environment in autonomous robot(link.1). The most important part is that the robot itself is able to localize and dynamically adapt to the changes in environment. The balloon calculates distance from the two different people to the balloon itself. After that, the balloon weights the different intensity of pull based on the different distances, which is called the calculation. As a result, instead of chasing the person who is closest, the behavior of the balloon will have different outcomes like moving toward the one person or moving toward the middle of two persons or wandering between the two persons. The balloons behavior is adapted from the different calculations within the environment it is in. It is interesting that by only applying one simple behavior rule, multiple meaningful interpretations of the balloon behavior is achieved .
Figure.2.2: The three simple rules for boids(Reynolds, 1987) and balloon only adapts rules from boids in choosing behaviors.
The second development of QUANTUMminus was exhibited to the public at Tate Britain. During this exhibition, we applied the second rule to balloon which is separation(link 2). Even though we still only have one balloon, the balloon could not come close many people in the space. Only by applying one more rule, many people in this experiment told me that they felt that the balloon was more like a cat pet rather than a dog.. Because even though cats sometimes comes close to you, they also sometimes set a distance in between and follow you around. The method of calculation is almost the same as last time project fair. The difference is in the choosing action not only cohesion but also separation. The size of balloon and space having which is environment having impact on perception of people would be also considerated by us(fig.2.3,fig.2.4).
Figure.2.3: Axonometric view of environment.
Figure.2.4: Plan and section at Tate Britain exhibition in July 2019.
Our project also considers as the implementation of Braitenberg Vehicles. In general a robot that produces a complex outcome does not need to be a complex robot. Therefore, Braitenberg Vehicles is the answer that exhibits complicated behavior robot building with simple principle and simple structure. The vehicles have limited capable of measuring its environment like our project in some way. It only measures variables including temperature or the amount of light or air pressure. The outcomes from sensors are directly driving motors. There is barely computation ability in vehicle, let alone the feeling or conscious in these vehicles. Some animations are created by us illustrating the Braitenberg Vehicles(link3). For instance, the vehicle in the experiment only has one sensor perceiving light and two dc motors moving in the same speed on each side that are assigned value from sensor. The behavior of vehicle is that vehicle hangs out in the dark and as it starts to see lights then it begins to drive a little bit faster. When it goes through the brightest part of the light and then slowing down and finally stopping when it gets to the dark again. Form personal perspective, the vehicle could be regard as a cockroach. When you enter a dark room and turn the lights on, they would hide on the dark because predators can see them being in the light. Therefore, the dark spot could be initiatively found by them.
The main point of Experiments in synthetic psychology is that synthesis is always easier than analysis or in other words creating something that acts complex is easier than analysing what looks like a complex system. Thus we can see that as an outsider looking at our vehicle it would be very easy to apply concepts to it about what it likes and dislikes based on the behavior that they observe. In this case, even though the algorithm and structure of vehicle is simple for us, the people could define the vehicle has intentional behavior. There are many vehicles like aggressive or fear or explorer or love. The love vehicle will initially drive towards the light. because the source is far away and dim but as it gets closer it will begin to slow down and enjoy the company of our light source. Once the light gets off to one side the motor speed mismatch will cause the vehicle to continue turning and ultimately begin speeding up in search of another friend. You would say that vehicle likes the source but actually it is always looking to explore the area for new friends like light source. It’s one true love when it’s far away it speeds up in search of it and once it spots the light the mismatch and motor speeds points the vehicle at its destination and as it gets closer it slows down and eventually comes to rest right at the source forever. The aggressive and fear and explorer vehicles are all have the similar algorithm. In this experiment Braitenberg describes how our simple machines could develop memory and the ability to predict and have foresight and develop ego from people perspective. From autonomous robot and Braitenberg Vehicles research, we realized that the simple behaviors we created contributing to diverse behaviors from people and then merging these various behaviors could result in complex perception(fig.2.5).
Figure.2.5: Simple rules to observed behaviors to merged behaviors(Braitenberg,1989).
According to Reynolds, the wandering behavior is a type of random steering which has a long structure. The steering direction of the next frame is related to the current steering direction, which means that the balloon can predict its future location. The wandering behavior we created is not a simple random motion, but a natural movement that can exhibit creature-like characteristic. Even though we generate wandering behavior using randomness, the mechanism predicting future location is adopted to keep the balloon moving in a less random-like way.
As mentioned above, wandering behavior is created by random goals or random desired velocity. We have created our own steering behavior through designing a vector system that describes the desired goal of the balloon.
When balloons have desired velocity, they will select actions by analyzing the elements in the whole system. Therefore, the balloons have essentially characteristic element of autonomous agents. The balloons perceive its environment and calculate all the elements within it , then select an action based on its desire. The most interesting part is that we can give different balloons different characteristics as well as different goals, creating an emergent behavior of a larger system that consists of both human and balloons.
Heider and Simmel discovered that when most people watch an animation of independently moving geometric shapes, they attribute intentional movement and goal-directed interactions to the shapes(fig.3.1). Nearly everyone who watches the animation, when asked to describe it, comes up with a fairly detailed “story” of what is happening. People developed a story from the movements of a bullying scenario that used a video of animated geometric shapes that seemingly interact. For Goncharova from tate modern exhibition, “The principle of movement is the same in the machine and in the living being.” From mentioned above, the emergent behavior is all about how people react to the behavior of the balloon. However, what I find interesting is how people feel about the movement of the balloon . is it possible for the balloon make people feel that the object is aggressive or friendly, like
Figure.3.1: Scenes from Heider and Simmel’s Experiment(Heider and Simmel,1944).
source : http://www.cell.com/cms/attachment/542632/3793833/gr3.jpg
It is intentional causality that we would put a reason or purpose to something often in a story or scenario. We could tell that from Heider and Simmel’s experiment, where an aggressive large triangle seems to be bullying a smaller triangle that is then continued with another frightened circle collaborating with the small triangle to setback the large triangle. There are many interactions between these agents around door. From Thinking, Fast and Slow, we’ve been found to have a desire to allocate human qualities to inanimate objects. there are two ways that we think. Both decisions that you made are based on the fast, automatic thought process, which Kahneman calls System 1 that is irrational and illogical thinking. The perception of emotion and intention is irresistible; only people afflicted by autism do not experience it. All this is entirely in your mind, of course. Your mind is ready and even eager to identify agents, assign them personality traits and specific intentions, and view their actions as expressing individual properties. Here again, “the evidence is that we are born prepared to make intentional attributions: infants under one year old identify bullies and victims” (Kahneman, 2011). After all, it is just a circle and two triangles changing coordinates. This animation helps us to demonstrate how readily we perceive social intention all around us. Moving shapes hit your eyes, but we see meaning and motives and emotion, all in the form of a social narrative. We can’t help but impose stories. “From time immemorial, people have watched the flight of birds, the movement of stars, the swaying of trees, and invented stories about them, interpreting them as having intention” (Eagleman, 2015). In UCL campus project fair, QUANTUMminus only havd one principle that is following people in a certain way. If we had more than two principles, people will definitely tell a story from the interaction. In Tate Britian, many people could already tell a story about the balloons because of two rules were applied to the balloon. From what we discussed above, the behaviors of an object could be constructed by a simple rule and a complex environment observed spontaneously by human beings.. After that, the personalities and patterns of behavior emerge from diverse behaviors and we could tell stories to form personalities(fig.3.2). A questionnaire was given to participants of our project in Tate Britain.
Figure.3.2: The process of observation from human being.
I received 12 feedbacks from the questionnaire where basically everyone graded more than 6 points in the questionss that address whether balloon itself has intension or feelings whether the participant felt uncomfortable when the balloon was approaching them. The result was that only one person felt uncomfortable when the balloon was approaching him. Even though most of the people felt that the balloon was like a pet for them, about 10 percent of the people felt that the balloon should move slower. . It was exciting to see how most of the participants felt that the balloon itself was positive and friendly, which means that the balloon carries favorable feelings and that people perceive the balloon as an object with feelings and intentions. Another interesting feedback about the balloon’s behavior revealed that a participant detected confusion and awkwardness from the balloon. The participant also perceived the balloon as a creature that is occupied with its own struggle. The brief of the project builds expectation, proposing the people could feel the balloon having intention. Because we added the second, there is more uncertainty in this experiment. This expectation naturally triggers our empathy for them, which means that every unpredictable, hesitant movement indicates confusion, as it would be in the case of living being, just like how it would be in the case of a living being.
There are many researches about the human robot interaction. Our balloon robot almost is the same size of people and our balloon doesn’t have gesture or posture, only pure primitive shape. That is why our robot experiment is unique. Maybe our balloon should understand and use proxemics in human robot interactions. “Proxemics is the study of distance and orientation and face-to-face encounters that have shown that people tend to stand closer to robots than to other people between 0.3 meter and 1.3 meters”( Mead, Atrash, Matari, 2013). In this range, the robots are able to automatically recognize human speech. However, the robots are unable to recognize human gestures because its cameras cannot see the entire body of the person. Thus the robots often fail during an autonomous social interaction with human. In our case there is no gesture or posture. How can we define proxemics in human interaction with movement of balloons that only has blurry orientation?
The concept of social distance is introduced in Georg Simmel’s article ‘The stranger’. In the article Simmel categorized social distance that exists in our social world as a vague classification of: wanderer, outsider, and stranger. “The stranger has elements of nearness and remoteness – he is attached, but not completely – while the social outcast is only remote” (Simmel, 1944). Simmel’s idea of ‘The stranger’ and the concept of social distance are linked through the categorization of humanity. In our daily scenario, different people have different definition of social distance which definitely has influence on our project. If people regard our balloon as human, the balloon needs to be not too close to humans because it can make people afraid. At the same time, the balloon could not stay too far away from people because of the attraction. The flocking algorithmic, where boids follow other boids and keeping distance from others, could be applied in our balloon project to make people feel interested yet comfortable. In this certain range, people could feel the balloon is interested and feel comfortable. However, a comfortable distance is subjective. For instance, Asian people usually stay in a closer distance compared to British people when they are talking to each other. Each person’s feeling of proxemics is based their culture(fig.4.1). Therefore, the Britain people in order to feel comfortable continuing keeping distance. Therefore, the Britain people in order to feel comfortable continuing keeping distance. From outside viewer, it seems like Asian people are chasing Britain people in our daily life. To make people feel comfortable and attracted by our balloons, proxemics should be taken into consideration.
Figure.4.1: Culture-based proxemics.
From the Disney Animation studies of The Illusion of Life includes 12 animation principles (1. Squash and Stretch 2. Anticipation3. Staging 4. Straight Ahead Action and Pose to Pose 5. Follow Through and Overlapping Action 6. Slow In and Slow Out 7. Arc 8. Secondary Action 9. Timing 10. Exaggeration 11. Solid drawing 12. Appeal) The study also discusses principles that affects how people see things and make objects more livable. First(1), The amount that an object squashes and stretches says something about its mass. The more squash and stretch, the softer the object. The most important principle related to our project is slow out and slow in(6). This principle refers to the way pretty much all movement starts slowly, build speed and finishes slowly. without slow in and slow out, things feel mechanical. That’s because robots are one of the only things that actually move their parts at a constant speed. To achieve various movements in QUANTUMminus, our balloon’s movement is controlled by a propeller moved by a dc motor that is connected to Optitrack sensors. Related to our new design strategy H-bridge, the propeller could turn different direction. In the first project fair, we used four transistors to control the speed of motors. If our balloons like traditional airplane without brake, then we could not achieve the effect of slow in and slow out and turn around. For instance, if the balloon only has one direction dc-motor, which make movement unrealistic like turning your back to someone. This is one of the most important principles to achieve lifelike motion. The other principle is called arcs(8). Very few organisms are capable of movements that have a mechanical in and out or up-and-down precision. Most living creatures will move in a circular path, well known as an Arc. Therefore, our balloons will not be straight line vehicles in the future so that I could become more user-friendly autonomous robots.
The speed of our balloon would affect how people perceive the balloon. The speed of up, down and forward and backward and rotate would be different related to physical structure and coding methods. How fast could the balloon move? Could we precisely control a balloon moving as we want it to move? From the exhibition in campus, we realized that the design had to change with a new physical model, we would be able to produce various kinds of movement and could more easily let people feel more story narrative movement that relates to Heider and Simmel’s theory of creating a story from human imagination. How would people react to these behaviors? Would they feel attracted or scared? According from the survey we conducted in Tate, almost 85 percent people felt that the balloon should move in a faster speed. Although it is impossible to literally control the balloons behavior, in order to control the balloon behavior more accurately, we changed to the motor from a drones motor to a pager motor. There are still many interesting surveies about balloons behavior, but we still do not know right now. Future development on the balloon’s behavior could be developed into serval behavior based on flocking algorithm. We have developed a couple of different behaviors based on the flocking algorithm. The first behavior is called love (fig.5.1), where the balloon approaches a human being slowly and then adapts its close distance to the human being. The second behavior is called ‘explored’(fig.5.1) where the balloon approaches a human being fast and then leaves them. Another behavior developed is called ‘fear’ (fig.5.2), which is similar to fear as living beings, where the balloon approaches the human but very slowly. Last we also explored ‘aggression’ (fig.5.2), where the balloon will approach a human being and keep a very close distance in between them. We also thought that a rotation behavior could be fascinating and engaging for people also prompting the experience as an
interactive space (fig.5.3).
Figure.5.1: the love and explored behavior.
Figure.5.2: the Fear and aggression behavior.
Figure.5.3: the rotation behavior.
The simplest form we adopted that keeping this project focusing on the movement of balloons. But
the features on the outside would effect how people recognize the movement of balloon. The alignment in the algorithm of flocking discuss these agents to steer towards average heading of neighbors. If we have a red spot or straight line on the front which is easily make people to perceive the balloon moving together. In these movement, the projection would be easier to make from people because of the orientation of balloon. For instance, the balloon is chasing you, the difference is the front side or reverse side facing you. The distinctive stories would be told from diverse orientation. In our daily life, if the person talks to you with his back, the reaction would be totally different. In the future investigation, we will have certain feature to indicate the more obviously orientation of balloon. At this moment, the movement of balloon is the only thing we consider to preserve the abstract of project.
Not only would people have projection on certain things, but also animal produce certain behavior respond to certain feature. For instance, the bird lover Christian mulac began flying with birds as part of a conservation project migrate with these birds in order to reintegrate them into nature(fig.6.1).He came up with the idea that he could teach a new bird generation to migrate to the protected area since these birds almost extinct as a result of human hunting. In this story, these geese recognize the wing of his aircraft and join the fly with him. Therefore, people could recognize certain features and have projection on them, so do the animal. The movement and form are the most important features for people to observe. Then movement could be noticed and make impression on human’s mind. As a result of thinking about this research, animal responds to certain feature whereas people would respond to certain movements and features.
Figure.6.1: flying with geese.
Zoological is a flock of autonomous, flying spheres that move collectively(fig.7.1). Algorithmically driven, the spheres react to their surroundings, sometimes, to people within their environment. This immerse show emerges from the exploration of human-machine interaction, demonstrating a new way of collective movement of inflated balloons – described by Random International as “organisms” are equipped with propellers, allowing them to buzz around the circular space, performing a live show . The drones’ path is determined by swarm algorithm making them move around in a flocking formation. The visitors are tracked by motion sensors, and as they wander around the space, the drones follow them. Basic behaviors of the balloons including curiosity, shyness and aggression are shown through the reaction with audiences.
Swarm algorithm they adopted are for better experience and performance. Swarm Intelligence is derived from the natural swarm behavior of animals which is defined as the collective behavior exhibited by the animals of same size, aggregating together to solve a problem which is essential for their survival. “SI can be defined as the emergent collective intelligence of groups of simple agents. Agents are analogous to the animals of the natural swarm”( Beni, Wang, 1993).
The term particles in place of points is from Particle swarm algorithm used the term particles in place of points in order to organize the idea of velocities cooperated with each particle. There are three simple rules including collision avoidance, velocity matching and flock centering for each particle to follow from flocking algorithm. “In addition to three rules, each particle has three inner attributes: it’s historical best position (local best position) and global best position of the swarm, and refers to those two positions whenever it moves to the next position“(Bonabeau,E.,Dorigo,M.,Theraulaz,G.,1999).These two positions stochastically determined the velocity of particle associated with another particle’s speed. Not until all the particles reach the most favorable solution or common goal, will they stop moving and updating their velocity. Hence, Particle swarm algorithm reaches the global maximum value and is not affected by the problem of local maxima. Swarm algorithm is mainly for performance not interacts with people rather than interaction.
Their project is similar in some ways but is on a larger scale and uses different technology. With similar approach but on a larger scale using different technology,they create a colony for people to experience, which is also our goal. However, these balloons always keep a distance from human being and do not directly react to human behavior. Besides, they are very noisy which is a big challenge for immersive atmosphere. .
Figure.7.1: Random International Zoological Drone Swarm.
In natural world, how do you tell thousands of agents like ant who are kind of stupid in a whole colony? The concept of nations emerges from human imagination just like the sense of colony emerges from ants. From a brief history of humankind (Harari, 2011), most of countries have fluid population changes and different symbol and border. There are many cities being abandoned and constructed all the time. However, these countries still exist because a lots of human interacting with each other. Even though we do not intend to create things, there are still many complex structures existing around us apart from nations. It seems that the society and political parties are constantly created by us that are fundamental properties for us.
According to the Social Self-Organization (Helbing, 2014), there are essentially two basic social coordination in the liner systems, which are the top-down approach from centralized authority and non-liner system. But within non-liner system is emerged from interaction of human on the local level that is much easier to maintain order compared to the centralized authority and this is what we call self-organization. It is easy for us to use self-organization as mechanism for coordination in unregulated large space. In our daily life we say hello to our neighbor when we occasionally meet them. We do not plan to meet them in advance, our local interactions are non-liner and also spontaneous. In the meantime, we do not have to ask for our neighbor’s permission to greet them, which is a distributed peer-to-peer interaction. Nowadays self-organization is becoming mainstream form of social interaction because of the increasing of information technology triggering distributed non-liner interaction. It is much more convenient to get local self-organization patterns of organization compared to centralized methods.
There are some correlations between the each agent’s state in the system to form distributed non-liner pattern. It is going to be total random if there is no correlation. Therefore, in self-organization, each person’s choice can be negative correlation or positive correlation to other’s choice. These agents would be coordinated themselves through feedback loop that plays the central role within the process of self-organization. Imagined that a person suddenly hears his favorite background music at a party and he get so excited then he starts dancing that is not supposed to happen. What happens next? It depends on the states of other person around him. Everyone will look at him, nonetheless not everyone feels he is wired. The positive feedback loop already created by us that helps others with propensity much effortless for dancing and joining him. The collective movement becomes more attractive as new individual joins. Now you will now be considered more normal if you are dancing rather standing aside because the dancing has already become a new attractor. Through this process of change driven by positive feedback loop, more people feel resistance to join the group. The languages and cultures in different regions are driven by certain attractor in these regions toward speaking a similar language and alike culture. Some small initial event coordinates with limited people and takes hold and then amplified into larger phenomena that forms an emergent self- organizing phenomena. Basically we cannot guarantee that which tiny behavior will take control of whole group behavior. If we have many balloons, then we will see swarm phenomena emerging from them.
Though the simple boids behavior could produce swarm phenomena. But the more interesting part is that we create small group behavior exhibits the different relationship or intentional behavior like Heider and Simmel.
There are also many attractors in the flock looking similar to human behavior. Because the evolutionary pressure from the environment, the birds found out it is much easier to survive to join a flock. There are three main reasons that contribute to bird to join the flock, which are protection from being hunted by predators, improving the gene pool they share and the increase of efficiency in searching for food as well as taking advantages for social and mating activities. Within an unregulated environment, dense distributed peer-to-peer interaction is required for self-organization in order to survive. Flocking of birds as a form of correlation between agent’s choices. If we have multiple balloons in virtual and physical world, then the swarm algorithm could be adapted afterward related to human-organization and boids-organization.
Emergent behavior refers to a bigger thing formed by many small thing that have convoluted properties that simplicity constituting complexity and emergence is everywhere. Flocking is an emerging property of birds. It created something new only by interaction within a huge amount of agents. This new behavior emerges from individuals’ interaction It turns out that more is different this different property is itself a new thing and that new thing can couple with other new things to repeat the process you can imagine this as layers stacked upon each other every layer made from more complex parts.
“This approach assumes that a flock is simply the result of the interaction between the behaviors of individual birds“(Reynolds, 1987).However, the emergent behavior that we created not only include balloon’s behavior but also include person’s behavior and is unpredictable. There was a couple fighting for the attention of the balloon in the project fair, which is hard to imagined before. This shows that balloons have the capability to get people’s attention. It also implies that the balloons may evoke a lot of spontaneous behaviors that is beyond our expectations. We cannot plan which balloon will encounter which person, and if we have many balloons and persons in a bigger picture, which means many operations emerge as a consequence from the interactions on fundamental level between many agents trying to maintain a vigorous and amazing structure and holding huge changes compared to only sum of its agents.
How can these ” stupid” thing with limited function be more than the sum of its part? Think about this, atoms form modules which then form proteins that make up cells afterward. After that, cells combine to form organs as parts of individuals. Finally, many individuals form a society. There are many creatures that have the same structure with human. Creatures emerge from non-living things that cooperate and combine each other, forming larger and larger systems. These things specialize in certain function and get feedback from each other forming complex organisms with impressively well-developed capability . There is no magic happening or a master giving command, just single cells communicating with neighbors, choosing action according to feedback they get. The most compelling thing is that our consciousness emerges from these cells, further scrutinizing, the chemical substances.
The flying balloons mechanism is inspired by vertical-moving drones with four motors that moving vertical. At first, two propellers moves horizontally and two propeller moves vertically(fig.9.1).. One of the vertical motors moves in clockwise vertically while the other one turns in anti-clockwise in order to making the sum of torque as zero. Well, a torque has the capability to rotate an object around fixed axis. The basic physical mechanism for motor moving horizontally is the same(fig.9.2). The problem is that we cannot control the balloon precisely as the two vertical propellers are controlling the height as well as rotation. Thus, we figured out a solution for this-the two vertical propellers only control the height, whereas horizontal propellers control rotation and forward-backward movement (fig.16). As what we mentioned above, we adopted H-bridge and pager motor to manage the behavior more properly. The drone motors are too powerful to control with h-bridge because of large current. Therefore, the pager motors are selected.
Figure.9.1: The mechanism of balloon.
Figure.9.2: The different way to control motors.
There are at least five materials being experimented in this project, which are foil balloons, weather balloons, PVC balloons, transparent 4D balloons and latex balloons. Since weather and PVC balloons are easier to explode, transparent 4D balloons and foil balloons are selected to make our robot. (fig.9.3).
Figure.9.3: The different material of balloon through experiment.
Optitrack is a motion capture system which provides high-performance optical tracking using infrared cameras and motion capture software. A variety of people ranging from film and games to sports training use optitrack equipment. In our case, the infrared cameras detect the makers on the balloons transforming data to unity through motive software and localize the balloon, which is the process of perceiving environment. Then, unity does the calculation which decides actions, and relevant data is sent to microcontroller that is installed at the bottom of the balloon.
At first, we used WIFI module ESP-8266 which works fine with transistor. However, the subsequently adopted H-bridge is not working properly with ESP-8266. Afterwards, both ESP-32 and Arduino wifi are all not working perfectly. Maybe raspberry pi zero or STM32 will be our final solution for microcontroller.
The biggest problem for us is to control the balloon to move accurately in the certain way. Our task is to build a flying balloon that can move from certain position to a designated destination. We need to figure out how fast the balloon is moving and rotating and then determine how long it takes for the balloon to reach the destination. There are so many unpredictable factors that are affecting the outcome of the balloons in the actual world. Therefore, it is essential to use feedback loop in order to minimize errors. Next, modification will be made according to the date from the optitrack cameras and feeding it back so that the system can make alteration based on now in a feedback system.
Because the first time the exhibition we directly jumped into feedback loop. However, there are a lot of problem about controlling balloons. In order to create the feedback loop, we should firstly design a proper non-feedback loop concerning the speeds, reference signals and the ultimate goal. Then we compare reference signal that is the destination to the location of the balloon and what you’re left with is the error between where you are and where you want to be. There are four calculations needed to be performed, which are movements in x, y, z axis and rotation (fig.9.4). Consequently in the case of the error would be the difference between the reference position and destination(fig.9.5). After that, the error signals are converted into the Pulse Width Modulation of motors which could help the balloons stay in constant level or destination.
Figure.9.4: Process of controlling balloons.
Figure.9.5: How to control motors in PID mechanism.
It is argued in this paper that autonomous balloon robot can be built by simple rules and simple structure with multiply possibilities with different rules. Before this research, it is almost impossible to create autonomous robot by ourselves. However, this experiment organizes the basic rules that i adopted for creating different behavior of balloon in return generating the delightful emergent behavior with people. At the same time, visitors tell stories and personalities of the balloons after experiencing our installation, demonstrating the phenomena of intentional causality. In this project, we realized that certain movements can trigger certain feelings, not to mention the influence of the form and size of balloons on human perception.
While there are an array of rules that we can employ to enhance the behavior of the balloons, the prime focus in the next few months should be the implementation of virtual reality design to create a more animated interaction between human and balloon based on the three simple rules from flocking system. At the same time, research about virtual balloon should be carefully carried out in order to improve the perception of intentional behavior of the balloons. All these aspects will help in creating an impactful evolving experience of autonomous balloon.
Beni, G., Wang, J., 1993. Swarm Intelligence in Cellular Robotic Systems, Robots and Biological Systems: Towards a New Bionics Springer, Berlin Heidelberg.
Bonabeau,E.,Dorigo,M.,Theraulaz,G.,1999.Swarm intelligence: from natural to artificial systems.
Braitenberg, V. Vehicles: Experiments in Synthetic Psychology . Bradford books: Psychology. MIT Press,1986.
- W. Reynolds. Steering behaviors for autonomous characters. In Proc. of Game Developers Conference, pages 763–782. Miller Freeman Game Group, San Francisco, CA, 1999. 
- W. Reynolds. Interaction with a group of autonomous characters. In Proc. of Game Developers Conference, pages 449–460. CMP Game Media Group, San Francisco, CA, 2000.
- W. Reynolds. Interaction with a group of autonomous characters. In Proc. of Game Developers Conference, pages 449–460. CMP Game Media Group, San Francisco, CA, 2000.
- W. Reynolds. (1987). Flocks, herds and schools: A distributed behavioral model. ACM SIGGRAPH Computer Graphics, 21(4), pp.25-34.
Dennett, D. and Braitenberg, V. (1986). Vehicles: Experiments in Synthetic Psychology. The Philosophical Review, 95(1), p.137.
Diana Nowacka, Nils Y. Hammerla, Chris Elsden, Thomas Pl¨otz, David Kirk. Diri – the actuated helium balloon: a study of autonomous behaviour in interfaces(2015)
Eagleman, D. (2016) The Brain: The Story of You, Canongate Books, Edinburgh.
Georg Simmel, “The Stranger,” in his Sociology (Glencoe, 111.: Free Press, 1969). See also Alfred Schutz, “The Stranger: An Essay in Social Psychology,” American Journal of Sociology 49 (1944).
Helbing, D. (2014). Social Self-Organization. Berlin: Springer Berlin.
Jafarinaimi, N., Forlizzi, J., Hurst, A., and Zimmerman, J. Breakaway: An ambient display designed to change human behavior. In Ext. Abstracts CHI’05 , ACM (2005), 1945–1948.
Jung, J., Bae, S.-H., and Kim, M.-S. Three case studies of ux with moving products. In Proc. of UbiComp ’13 , ACM (2013), 509–518.
Kahneman, D. and Egan, P. (2011). Thinking, fast and slow. New York: Random House Audio.
Liu, S. (n.d.). Creating autonomous vehicle systems.(2018)
Metzinger, T. (2011). The ego tunnel. [New York]: RHYW.
Nowacka, D., and Kirk, D. Tangible autonomous interfaces (tais): Exploring autonomous behaviours intuis. In Proc. of TEI ’14 , ACM (2014), 1–8.
Ross Mead, Amin Atrash, and Maja J Matari. “Automated Proxemic Feature Extraction and Behavior Recognition: Applications in Human-Robot Interaction,” International Journal of Social Robotics, v.12369, 2013, p. 1-12.
Sapiens: Yuval Noah Harari(2011). A Brief History of Humankind
Thomas, F., Johnston, O. and Thomas, F. (1995). The illusion of life. New York: Hyperion.
Towaha, S. (2018). Building Smart Drones with ESP8266 and Arduino. Birmingham: Packt Publishing.
Tsuchida, S., Takemori, T., Terada, T. and Tsukamoto, M. (2017). Mimebot: spherical robot visually imitating a rolling sphere. International Journal of Pervasive Computing and Communications, 13(1), pp.92-111.
Welsby, J. and Melhuish, C. (2001). Autonomous Minimalist Following In Three Dimensions: A Study with Small-Scale Dirigibles.
 Optitrack is a motion capture system