Register now After registration you will be able to apply for this opportunity online.
Urban Navigation for Wheeled-Legged Robots
Traditional legged robots are capable of traversing challenging terrain but lack energy efficiency and speed when compared to wheeled systems. We propose a novel optimization/learning framework that enables a legged robot equipped with powered wheels to perform hybrid locomotion.
Recent research in robotic locomotion has shown a great variety of locomotion strategies. A large part of this work has been focused on traditional walking maneuvers, where the feet are assumed to remain stationary when in contact with the environment. However, comparisons between wheeled and legged robots have shown that the former is much superior on flat terrain compared to legged robots in terms of energy efficiency and speed. In this work, we focus on hybrid locomotion strategies for a wheeled-legged robot and as such, we combine the advantages of both locomotion domains in one locomotion framework. In addition, we exploit the performance of the novel locomotion framework in an unstructured environment.
In recent work, we showed highly agile and complex motions of our roller-walking robot, ANYmal. The robot can locomote on flat terrain up to 3 m/s and over overcome challenging obstacles with up to 1.5 m/s. In this project, we would like to go one step further and make this robot fully autonomous by applying novel navigation techniques.
Therefore, possible subtopics of this work can be summarized as follows:
- Navigation
- Perception
- Localization
- Path planning
- Machine learning
- Optimal-control
See the videos below:
See videos and information here: https://www.swiss-mile.com/
See all published work with ANYmal on wheels:
https://www.markobjelonic.com/research/
Recent research in robotic locomotion has shown a great variety of locomotion strategies. A large part of this work has been focused on traditional walking maneuvers, where the feet are assumed to remain stationary when in contact with the environment. However, comparisons between wheeled and legged robots have shown that the former is much superior on flat terrain compared to legged robots in terms of energy efficiency and speed. In this work, we focus on hybrid locomotion strategies for a wheeled-legged robot and as such, we combine the advantages of both locomotion domains in one locomotion framework. In addition, we exploit the performance of the novel locomotion framework in an unstructured environment.
In recent work, we showed highly agile and complex motions of our roller-walking robot, ANYmal. The robot can locomote on flat terrain up to 3 m/s and over overcome challenging obstacles with up to 1.5 m/s. In this project, we would like to go one step further and make this robot fully autonomous by applying novel navigation techniques.
Therefore, possible subtopics of this work can be summarized as follows:
Please note that due to the limited number of positions, the selection is highly competitive. Applicants should have experience in some of the following topics:
- C++ programming
- Machine learning
- Optimization theory
- Reinforcement learning
- Navigation
- Robotics
Please note that due to the limited number of positions, the selection is highly competitive. Applicants should have experience in some of the following topics:
- C++ programming - Machine learning - Optimization theory - Reinforcement learning - Navigation - Robotics
Your application should include a brief motivational statement, your transcript of records and your CV.
Your application should include a brief motivational statement, your transcript of records and your CV.
