Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
ANYmal tails for stability during discontinuous locomotion
Animal tails serve all manner of functions, ranging from locomotion stability to self-defense. In legged robots, tails may enable a new generation of stable and agile machines that approach the performance of animals. In this project, students will perform simulation studies to determine if tails improve robotic locomotion of the quadruped robot, ANYmal, over discontinuous terrain (gaps, steep steps, etc.; see [1] for an example tail parametrization and modeling procedure).
Legged animals have used tails for many purposes: as a counterbalance to maintain stability [2], for rapid maneuvering [3], and even as a weapon [4]. Much prior work has explored the potential of robotic tails in the domain of locomotion. However, studies have largely focused on i) hardware experiments with tail designs based on heuristic selection or rote biomimicry [5]; ii) simulated control of crude tail approximations, such as point-mass systems [6]. In this simulation-based project, we hope to effectively model and search the design space of tails—including parameters like morphology, mass distribution, stiffness, and number of actively controlled degrees of freedom—to assess what factors might be relevant to the locomotion performance of tailed quadruped robots. We also hope to gain insights about the biological function of tails over challenging and discontinuous terrain: Do high-performing tails in simulation mirror those we see in nature? In what cases do tails help versus diminish performance? This project will be co-supervised with Prof. Dr. Ardian Jusufi (EMPA Soft Kinetic Group leader https://www.empa.ch/web/juar ).
References:
[1] Fukushima et al. 2021. https://academic.oup.com/icb/article/61/2/589/6261076#304011199
[2] Jusufi et al. 2008. https://www.pnas.org/doi/epdf/10.1073/pnas.0711944105
[3] Patel et al. 2015. https://link.springer.com/chapter/10.1007/978-3-319-06773-5_10
[4] Arbour et al. 2019. https://anatomypubs.onlinelibrary.wiley.com/doi/full/10.1002/ar.24093
[5] Chang-Siu et al. 2011. https://ieeexplore.ieee.org/document/6094658
[6] Yang et al. https://arxiv.org/pdf/2303.04781.pdf
*Note: in your application, attach your cv and transcripts to receive full consideration for the position. A cover letter of motivation will also help us decide if you are a good fit.
Legged animals have used tails for many purposes: as a counterbalance to maintain stability [2], for rapid maneuvering [3], and even as a weapon [4]. Much prior work has explored the potential of robotic tails in the domain of locomotion. However, studies have largely focused on i) hardware experiments with tail designs based on heuristic selection or rote biomimicry [5]; ii) simulated control of crude tail approximations, such as point-mass systems [6]. In this simulation-based project, we hope to effectively model and search the design space of tails—including parameters like morphology, mass distribution, stiffness, and number of actively controlled degrees of freedom—to assess what factors might be relevant to the locomotion performance of tailed quadruped robots. We also hope to gain insights about the biological function of tails over challenging and discontinuous terrain: Do high-performing tails in simulation mirror those we see in nature? In what cases do tails help versus diminish performance? This project will be co-supervised with Prof. Dr. Ardian Jusufi (EMPA Soft Kinetic Group leader https://www.empa.ch/web/juar ).
References:
[1] Fukushima et al. 2021. https://academic.oup.com/icb/article/61/2/589/6261076#304011199
[2] Jusufi et al. 2008. https://www.pnas.org/doi/epdf/10.1073/pnas.0711944105
[3] Patel et al. 2015. https://link.springer.com/chapter/10.1007/978-3-319-06773-5_10
[4] Arbour et al. 2019. https://anatomypubs.onlinelibrary.wiley.com/doi/full/10.1002/ar.24093
[5] Chang-Siu et al. 2011. https://ieeexplore.ieee.org/document/6094658
[6] Yang et al. https://arxiv.org/pdf/2303.04781.pdf
*Note: in your application, attach your cv and transcripts to receive full consideration for the position. A cover letter of motivation will also help us decide if you are a good fit.
- Literature research on legged animals with tails and robots with tails
- Simulation of quadruped robot with various tail types
- Implementation of reinforcement learning algorithm or evolutionary algorithm to converge on ‘optimal’ tail designs for challenging locomotion scenarios
- Literature research on legged animals with tails and robots with tails - Simulation of quadruped robot with various tail types - Implementation of reinforcement learning algorithm or evolutionary algorithm to converge on ‘optimal’ tail designs for challenging locomotion scenarios
- Excellent knowledge of programming and robot simulation
- Independence and creativity
- Preference: coding skills in C++ and familiarity with reinforcement learning
- Excellent knowledge of programming and robot simulation - Independence and creativity - Preference: coding skills in C++ and familiarity with reinforcement learning
Robert Baines
rbaines@ethz.ch
Yuntao Ma
mayun@ethz.ch
Ardian Jusufi
ardian.jusufi@uzh.ch