2084: Playing frisbee with robodogs on the beach
The new robot dog Raibo can run at up to 10 kph or 7 mph on a variety of surfaces, from asphalt to grass to sand.
I was scrolling through my news sites when I saw an article on a brand new robot developed in Korea which could run on sand at 10kph. Now if you’re from a town close to the coast, you know, running on sand is not that easy. Your footing is inherently unfirm, and you have to make constant adjustments, and so to have a robot that could do that reliably is absurdly hard. But as the following video shows, they managed to make it a reality:
It is quite freaky to see though. The associated paper is called learning quadrupedal locomotion on deformable terrain and its quite interesting to see what they did: basically they built a physics simulation that could be parameterized to represent a vast array of different surfaces, and by varying the training data among these surfaces, they could train a control module capable of dealing with a large variety of surfaces.
It reminds me quite a bit of the way in which Robotics Transformer 1 was trained. That was trained on a variety of data as well, a combination of simulation and real data, for a variety of different robots, and with this variety, it achieved impressive real world robustness. Now we just need to bring the two together: Have Robotics Transformer 1 handle the natural language and decision making capabilities, and use the techniques in this paper to handle movement across a variety of different surfaces at speed. The ideas behind varying the physics simulation and the construction of the control module could surely be generalized to produce a faster and more competent robot grabber, or whatever other function is necessary. It can’t be too far away that we’ll have robodogs with arms cleaning up around the house, who can respond to complicated natural language instructors and even seem like they react and feel emotion. What a future to be moving towards!
Sources
Suyoung Choi, Gwanghyeon Ji, Jeongsoo Park, Hyeongjun Kim, Juhyeok Mun, Jeong Hyun Lee, Jemin Hwangbo*. "Learning quadrupedal locomotion on deformable terrain", Science Robotics [https://www.science.org/doi/10.1126/scirobotics.ade2256]
Anthony Brohan and Noah Brown and Justice Carbajal and Yevgen Chebotar and Joseph Dabis and Chelsea Finn and Keerthana Gopalakrishnan and Karol Hausman and Alex Herzog and Jasmine Hsu and Julian Ibarz and Brian Ichter and Alex Irpan and Tomas Jackson and Sally Jesmonth and Nikhil Joshi and Ryan Julian and Dmitry Kalashnikov and Yuheng Kuang and Isabel Leal and Kuang-Huei Lee and Sergey Levine and Yao Lu and Utsav Malla and Deeksha Manjunath and Igor Mordatch and Ofir Nachum and Carolina Parada and Jodilyn Peralta and Emily Perez and Karl Pertsch and Jornell Quiambao and Kanishka Rao and Michael Ryoo and Grecia Salazar and Pannag Sanketi and Kevin Sayed and Jaspiar Singh and Sumedh Sontakke and Austin Stone and Clayton Tan and Huong Tran and Vincent Vanhoucke and Steve Vega and Quan Vuong and Fei Xia and Ted Xiao and Peng Xu and Sichun Xu and Tianhe Yu and Brianna Zitkovich, RT-1: Robotics Transformer for Real-World Control at Scale, https://robotics-transformer.github.io/
O.O really cool stuff m8 !