Auke Ijspeert and his team at the Laboratory of Biorobotics ( BioRob ) in EPFL’s Faculty of Engineering had already tested their bio-informed robots in the wild. But this was more for demonstration purposes than for scientific rigor. Robot function tests are usually carried out in the laboratory, for example using X-ray video to compare robot movements with those of the animals that inspired their design.
Things changed in November 2015: Auke Ijspeert and his colleagues received a request from BBC producers to create two realistic robots, one to mimic a crocodile and the other, a monitor lizard. Both species live on the banks of the Nile in Uganda. BioRob’s mission is to design and build, in less than a month, robots concealing cameras and capable of discreetly integrating into this environment in order to film the nesting behavior and interactions of these reptiles.
Technical adaptations
The request seemed straightforward enough, and the scientists relied on their experience in creating robots with a spread-out posture, such as Pleurobot and Orobot. But the first challenge arose when it came to finding a balance between form and function: the robots developed as part of the Krock platform - SpyCroc and SpyLizard - had to blend in with real crocodiles and varans to film their interactions, so a higher percentage of their weight had to be allocated to the cameras and hyper-realistic skin.
"Part of the design process involved anticipating what might happen later and simplifying the design as much as possible to make it easier to repair the robots in the field, where access to specialized parts and equipment is limited," explains Kamilo Melo, former postdoctoral researcher at BioRob and now director of biorobotics company KM-RoBoTa. To achieve this, scientists have turned to low-cost components that are easy to exchange or replace.
In Uganda, working conditions in the field pose unexpected challenges. With the mercury at 38 degrees, the temperature inside the robots climbed to 80 degrees, causing them to overheat and shut down. The scientists therefore had to work quickly before the daytime temperature rose, and get around the problem by, for example, running the robots for short periods interspersed with cool-down periods. They had to simplify the robot design as much as possible to reduce the number of connecting parts, as more joints meant more entry points for sand, dust and moisture. What seemed to be an asset of the Krock design - such as structural rigidity - turned out to be a problem, as the rough terrain would simply have caused the rigid components to break.
BioRob has published the lessons learned as a freely accessible research and methodology resource in the journal Science Robotics. The scientists hope that their experience, coupled with design specifications using simple, robust and commonly available components, will help others to replicate their platform for their own projects.
A more efficient biorobot
Building on their experience in Africa, the scientists have developed a new version of the Krock platform, Krock-2, which is more robust, flexible and waterproof. Requiring fewer elaborate camouflage elements such as a realistic latex skin, the improved robot has great potential for rescue and disaster response.The experiment has also opened up new avenues of research in the laboratory. "The development of a tactile skin with sensors capable of detecting the forces of interaction with the environment is an important subject that integrates pure robotics and neuroscience," says Auke Ijspeert. "In robotics in general, we’re very good at reproducing proprioception, but very bad at reproducing all the senses we have in our skin, such as heat and touch. This technology is still very complex, and we want to integrate it into our salamander robots."
For the two researchers, the technical improvements made to the Krock platform based on the field tests are more than just a bonus. They want to use the experience gained in Uganda to further develop bioinformed robots as scientific tools, for example in robotic paleontology to understand the locomotion of extinct species such as dinosaurs. Although bones and fossils can be used to create animations and study kinematics to understand the dynamic movements of dinosaurs, a physical model must be built that is subject to the same laws of physics as animals of the past.
"Thanks to the freely accessible contributions we’ve made as part of this study, we hope to make these platforms more affordable, but still sufficiently accurate for scientific purposes."
