Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Foothold Generation for blind Locomotion in rough Terrain
An important task in legged locomotion is to place the feed on locations that provide sufficient support to maintain stability. In addition, the robot should achieve a desired travelling velocity while ensuring kinematic feasibility of the footholds.
For our locomotion framework [1] for legged robots we look for a new method to generate desired footholds (i.e., point locations on the ground to place the end-effectors). The foothold module represents a crucial part in the locomotion pipeline since it is directly used to formulate a dynamic stability criterion for the torso pose optimization.
The student has to choice between one of the following directions:
1. A modular approach that computes a set of footholds independent of the pose optimizer.
2. A module based approach that combines pose and foothold optimization into one framework.
This project will establish the first milestone for a robust vision-based foothold generator that will be employed on ANYmal in the upcoming DARPA challenges.
[1] C. D. Bellicoso, F. Jenelten, C. Gehring and M. Hutter, "Dynamic Locomotion Through Online Nonlinear Motion Optimization for Quadrupedal Robots," in IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 2261-2268, July 2018
For our locomotion framework [1] for legged robots we look for a new method to generate desired footholds (i.e., point locations on the ground to place the end-effectors). The foothold module represents a crucial part in the locomotion pipeline since it is directly used to formulate a dynamic stability criterion for the torso pose optimization.
The student has to choice between one of the following directions: 1. A modular approach that computes a set of footholds independent of the pose optimizer. 2. A module based approach that combines pose and foothold optimization into one framework.
This project will establish the first milestone for a robust vision-based foothold generator that will be employed on ANYmal in the upcoming DARPA challenges.
[1] C. D. Bellicoso, F. Jenelten, C. Gehring and M. Hutter, "Dynamic Locomotion Through Online Nonlinear Motion Optimization for Quadrupedal Robots," in IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 2261-2268, July 2018
- Review relevant packages
- Review literature
- Implement new foothold generator
- Experimental verification with ANYmal on rough terrain
- Review relevant packages - Review literature - Implement new foothold generator - Experimental verification with ANYmal on rough terrain
- good knowledge in C++ programming
- online optimization (MPC, QP, SQP)
- experience with ROS is a plus
- high motivation
- good knowledge in C++ programming - online optimization (MPC, QP, SQP) - experience with ROS is a plus - high motivation
For further informations please contact Fabian Jenelten (fabian.jenelten@mavt.ethz.ch).
For further informations please contact Fabian Jenelten (fabian.jenelten@mavt.ethz.ch).