Why do we build robot animals?
When people ask about my grad student research, they’re sometimes taken aback by my response: “I build robot lobsters.” Upon hearing this far out statement I’m guessing most people envision a mix between Zoidberg, the lobster alien from Futurama, and Big Dog, YouTube’s most popular animal-like robot. I try to give more details so people understand what I actually do but that can be difficult when explaining the science over the roar of the crowd at Fenway or in the middle of a weekly pub trivia quiz.
Wherever I am, one of the main follow up questions I get is:
why on Earth would someone want to build a robot lobster?
Today I’ll tell you why we build biomimetic robots, robots engineered to mimic animals. There are three different but equally important reasons:
1. We can build better robots by learning how animals solve problems.
Animals have evolved for hundreds of millions of years and are very well suited to the environments they inhabit.* If we want to build a robot that can operate in a particular environment, like the ocean floor, why not copy a solution that nature has come up with, like the lobster? We want our robots to be able to overcome any challenges they face, and animals have already figured out how to do that. Rarely do we see animals stuck in the wild. If an animal were to get stuck, it would wiggle and squirm its way to freedom. Traditional autonomous robots can’t really do this as their behaviors have to be programmed ahead of time. In our robot animals we are trying to mimic the way the animal nervous system produces this wiggling and squirming. If our robots can produce new solutions to unforeseen circumstances like animals can, we’ll have some pretty capable robots on our hands.
2. We can learn about the animals by using the robot as a hypothesis-testing platform.
Nervous systems are complicated. The human brain has billions and billions of neurons. Even the relatively simple nervous system of the lobster has over 30,000 neurons. Understanding the connections (synapses) between all these neurons is a tremendous challenge and it still isn’t enough to explain the activity of a nervous system (this leads to a debate about the ongoing efforts to map the human brain which I’ll save for another day). Substances called neuromodulators can change the activity patterns of neurons and change the synaptic connectivity between neurons. On top of all of that, much of what we know about the nervous system activity in lobsters and other animals has been learned from what are called isolated preparations. This means that neuroscientists remove the nervous system from the animal and then conduct recordings of electrical activity (the language of neurons) in a Petri dish. Networks of neurons often act very differently in a tabletop dish of saline compared to what they do in a freely behaving animal. What I’m trying to tell you is that our understanding of nervous systems, even in supposedly ‘simple’ organisms, is quite limited.
We adopt an approach that uses robots to learn about biology called biorobotics. Based on the biological research evidence available, we come up with a hypothesis for how the nervous system of the lobster controls a particular behavior (e.g. the motor response to visual movement). The robot lobster gives us a platform to test out this hypothesis. We can run comparative testing between animal and robot in identical situations to see how their behaviors compare. If the response of the robot is different than that of the animal, we know that our hypothesis is deficient and we have an idea of where to focus further biological experiments. We can also validate some of the biological findings through this robot testing.
3. Robot animals are intriguing, charismatic and cute (depending on your tastes) and inspire students and the general public to engage in science.
Robot animals are captivating and quickly garner interest from people of all sorts: from students to neighbors to grandparents. Having such a visually interesting embodiment of my research gets others engaged in the science by asking questions (such as those posed in this post). One major goal of mine is to educate and inspire young scientists and robot animals are a great way to reach a younger audience. Our lab values educational outreach and sharing our work with others whenever possible and I hope that these efforts push more students to pursue careers in science and technology.
I hope this post has given a better understanding of why we build robot lobsters. Now you may be wondering why we chose to mimic lobsters out of all the animals out there. I’ll have to answer that question in my next post.
*Notice I didn’t say that animals are ‘optimized’ or ‘perfectly suited’ to their environments. It’s important to note that animals always continue to evolve over each generation. While natural selection usually pushes evolution towards improving an animal’s characteristics for a given environment, the randomness of the process sometimes leads to unexpected outcomes. Through evolution, animals can change so much that their preferred niche (their location and ecological role) changes. The most prominent evidence for why we can’t use the word ‘optimized’ to explain the result of evolution is that of vestigial organs. Why do ostriches have wings? Or why do we humans have appendixes? These are remnants of previous evolutionary changes that no longer provide much function in the present day and could be described as less than optimal.