Professor Marc Bohlen and his chickens want respect. From a robot.
The University at Buffalo professor of media study set out to get that respect in a robot-chicken interaction experiment titled “Advanced Perception.”
Respect, for Bohlen, meant that the chickens could live in a six by six-foot cage with a robot and not show fear. One of his goals was to have the chickens go about their daily routine with a robot wandering about with them.
Bohlen wanted to show that humans could build something motivated and active that didn’t present harm to chickens. Chickens are, to him, often overlooked. Once he had made this statement, Bohlen hoped that others would take the hint and think about machines that didn’t necessarily hurt animals.
Bohlen is the keeper of the media robotics lab at UB. This is a term he invented to stress that everyone is working together. “I could have called myself the Lab Director, but I like to think of myself as being more of a gardener,” he says.
Bohlen is interested in finding new ways to use automation and robotics. He wants to explore the connection between art and robotics, a theme which is shown through his broad education. He has received Masters degrees in art, art history, and robotics, as well as a Bachelor’s in electrical engineering.
The different disciplines in Bohlen’s background all come to surface in his work. Projects that he has either completed since the chicken experiment in 1999 or are now in progress include a machine that communicates by whistling, robots that speak in different dialects, and a robotic boat.
“Task completion is not the only place that we can have automation,” he says. “It can go other places, too, and sometimes strange, weird, wonderful, and dumb things can happen.” With “Advanced Perception” Bohlen wanted to prove that we could build machines that would move and accomplish their own goals without frightening or harming animals.
To conduct the experiment, he placed three Rhode Island Red hens in a closed environment in his own small studio space. As each chicken moved around the pen, a camera above it would track the movements and send the information to a computer.
The computer sent directions to a small hockey puck-shaped robot within the spacious chicken cage. The robot would snap to life and, on its black treaded wheels, back away from the chicken, allowing it to pass.
Bohlen built and tested this project in 1999 in a lab at Carnegie Mellon University. It is an attempt to “find a new way humans can create systems that aren’t just selfish,” Bohlen says. Bohlen’s concern is that manufacturing machines serve only our ends, while chickens are slighted.
For Bohlen, chickens are the most maligned animals in industry. “The Industrial Revolution was supposed to make the world better,” Bohlen says. “For chickens, it didn’t work out that way.”
He was able to program the robot in “Advanced Perception” to back out of the way of the chickens. It made itself known, by jittering its motor, before moving and alerting them. The robot wandered the cage, much the way the chickens did, without frightening them.
Though he was successful in programming the machine, he was not pleased with the responses to his work. “The awful thing is that the agriculture industry heard about it and approached me,” Bohlen says. They wanted him to make a robot that could lure chickens in from their cages to be killed.
When he presents this work, the quick reaction of observers is often that his robot could be used to benefit the meat industry.
Bohlen’s intention was to prove that “we can build something that treats creatures well.” The invitation from the agriculture industry to use his machine for meat production is the opposite of this intention.
Bohlen doesn’t eat meat and claims that creating a machine to kill chickens is just about the last thing he would do. His artistic statement, that machines ought to be built to work with animals, may not have swayed his observers. Bohlen’s conclusion is that “people are not ready to build machines that treat other creatures with respect.”
Bohlen has recently been working on robotic communication systems. The first system communicates with a language that is easily identifiable and interpretable: whistling.
The Universal Whistling Machine (UWM) is a small, white, tent-looking device with a speaker and a microphone. There are still two of these devices on the shelf of his office in the Center for the Arts. The UWM is able to detect passersby with its built-in camera and then, lure them into a whistling exchange by emitting an introduction. To introduce itself, the machine whistles a pattern that it remembers from someone else.
The machine was created with the participation of UB student JT Rinker, a music composition student and Ph. D candidate. According to Bohlen, whistling is a way of speaking to machines on a middle ground between programming languages and human speech.
Stuart Shapiro is a UB professor of computer science and a specialist in computational linguistics. Shapiro claims that communicating through whistling may represent what he calls a “symbol-grounding problem.”
“I guess the question is, ‘how does meaning get into it?’” he says. “You have to indicate what you mean by those whistles.”
Shapiro admits that the utility of the whistling machine might be the wrong question. He is more interested in the machine as being capable of posing an artistic question.
Bohlen is aware of the limitations of his machine’s utility. “There’s no utlity like there’s no utility to wearing a red scarf,” he admits. “You wear it because you like it.” Though it may not communicate meaning, the machine can often provoke response from a human, a bird, or even another whistling machine.
Bohlen is currently working on a newer communication system. The work is a text-to-speech system. The first such system was created by Ray Kurzweil in 1976. It looked something like a Xerox machine and was used in libraries to communicate text to individuals who were visually impaired.
According to Shapiro, the system was “fantastically useful, regardless of pronunciation.” The system’s speech was not perfect, but, according to Shapiro, it was good enough.
The difference in Bohlen’s text-to-speech system and others such as Kurzweil’s is that he intends to make it more believable by being less robotic. The voices will have foreign accents that make them difficult to understand. According to Bohlen, making these machines with accents will “push the notion of believability one step farther.” Bohlen’s proposition suggests that being less than perfect makes a machine more believably alive—what he calls “hypernatural.”
These text-to-speech systems are a preliminary step in Bohlen’s development of an autonomous glass bottom boat. In the proposed project, six participants would board a small glass bottom boat in a shallow lagoon in Florida. The boat will have no captain except for on-board navigation powered by a global positioning system.
Several digital video cameras underwater will record the terrain and its inhabitants. Bohlen’s new text-to-speech systems will attempt to narrate what’s going on below the boat. Cameras will work to recognize wildlife.
Venu Govindaraju is a UB professor of computer science. He specializes in pattern recognition systems. Govindaraju recognizes how difficult it is to make robotic vision and recognition. According to him, robots are not yet able to walk into an area as simple as an office space and distinguish objects. He says that it is much easier for a robot to recognize objects in a “limited, constrained environment.”
Bohlen knows it will be difficult to distinguish objects underwater, in an environment that is neither limited nor constrained. “There will be mistakes,” he admits, though he hopes to incorporate the mistakes into his work by replacing images that the robots cannot recognize with invented stories about the lives of the fish beneath them. And when the water gets murky and the cameras don’t see anything at all, the screens will display things that aren’t there: shopping carts and derelict tires, unreal fish and computer consoles.
Bohlen spent winter break in Florida scuba diving for footage for the boat’s displays and meeting with possible partners on the project.
Junda Lin, a marine biologist at the Florida Institute of Technology, met with Bohlen for a short talk about the project. He admits that underwater visualization is a tricky thing. According to Lin, any advances made in Bohlen’s work could be very helpful in marine biology as well.
“Certainly, there’s a lot of potential;” he says, “there’s a lot of potential for environmental issues.”
Though the logistics are nightmarish and the project is expected to take three years and $100,000, Bohlen says, “I’m sure I can pull it off.”
He sees the boat as an observation post. From the deck, passengers will be able to view underwater life and hear tales about the fish—some of which will not be true. Bohlen considers the stories tall tales that we would like to believe although our better inclination is to deny them.
However, Bohlen firmly denies that the tales will be lies. “It’s not an attempt to mislead,” he says. “People are curious, but afraid of the beach; they want stories.” According to him, the boat will raise issues about the clash of comfort and the desire to experience the awful force of nature. “People want to protect, participate [in], and stay away [from nature],” Bohlen says.
The necessity of these projects is not always obvious, and sometimes there simply isn’t any. According to Bohlen, “We play sports or play bridge or go out drinking. These are all things that aren’t necessary, but we spend half our lives doing them. The same goes for machines.”
