MacCurdy and collaborators advance research in genetic and evolutionary computation
Assistant Professor Robert MacCurdy and his collaborators have won the for their outstanding contributions in the field of genetic and evolutionary computation.
The award recognizes up to three papers a year that were published in the 10 years earlier and have amassed a high level of citations and deemed to be seminal. Their paper, titled 鈥,鈥 was the only paper to receive the award in 2023.
MacCurdy coauthored the paper along with , and , the latter being the head of the , where MacCurdy and his collaborators met and did the work.
The paper was inspired by the created by ; in his research, Sims demonstrated that computational evolution can produce morphologies that resemble natural organisms, but the potential for increasingly complex and natural morphologies hit a ceiling. It was hypothesized that the limitation in morphological types was due to the rigidity of the materials used in the design space and the direct encoding.
Addressing these problems in their paper, MacCurdy and his collaborators demonstrated how computational evolution can be pushed further through the creation of soft robots and the use of generative, evolutionary-based encoding that wielded the power of multi-objective optimization.
鈥淲hen you鈥檙e trying to solve a design problem, it鈥檚 smart to show some humility because you don鈥檛 ever fully know the true nature of the problem,鈥 said MacCurdy, 鈥渟o it鈥檚 appealing to use a multi-objective design framework that gives you a whole population of very different solutions to that set of design objectives.鈥
Using their novel approach, MacCurdy and his collaborators were able to create a set of virtual robots whose locomotion resembled animals found in the natural world but also creatures whose gait was wildly idiosyncratic and unique.
鈥淪ome robots galloped like a horse. Others had the running gait of a dog or rolled along like a walrus,鈥 MacCurdy said. 鈥淚 think these designs were able to capture people鈥檚 imagination, while also motivating the use of generative algorithms and multi-objective optimization to solve challenging design problems, and that鈥檚 why the paper continues to garner citations and serve as an inspiration for others.鈥
A video of the work has garnered hundreds of thousands of views. Watch it here:
[video:youtu.be/z9ptOeByLA4?si=cNA1CG5olCEsncze]
The impact of the paper has received recognition at several conferences, while the SIGEVO Impact Award cements its importance in the robotics community. The real-life applications of the paper have furthered the study of evolutionary biology and the design of soft robots that can move in the real world.