Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Reactive Obstacle Avoidance for Mobile Manipulation
The goal of this project is to achieve fast and reactive obstacle avoidance with a real mobile manipulator.
Keywords: Robotics, Mobile Manipulation, Motion Planning, Perception
Manipulators have to work in dynamic environments, where the location of obstacles can quickly change. Reacting to these changes requires a tight coupling between the map of the environment and the planner/controller in order to ensure safe and efficient motions of the manipulator arm.
The goal of the project is to investigate environment and obstacle representations [2,3] that are suited to be used within fast and reactive local planners, such as Riemannian Motion Policies [1].
References:
[1] Riemannian Motion Policies, Ratliff et al., 2018
[2] Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-board MAV planning, Oleynikova et al., 2017
[3] FCL: A general purpose library for collision and proximity queries, Pan et al., 2012
Manipulators have to work in dynamic environments, where the location of obstacles can quickly change. Reacting to these changes requires a tight coupling between the map of the environment and the planner/controller in order to ensure safe and efficient motions of the manipulator arm. The goal of the project is to investigate environment and obstacle representations [2,3] that are suited to be used within fast and reactive local planners, such as Riemannian Motion Policies [1].
References: [1] Riemannian Motion Policies, Ratliff et al., 2018 [2] Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-board MAV planning, Oleynikova et al., 2017 [3] FCL: A general purpose library for collision and proximity queries, Pan et al., 2012
1. Literature review on riemannian motion policies and obstacle representations.
2. Integrate real-sensor data into RMP pipeline
3. Evaluate performance in simulation and on real robots.
1. Literature review on riemannian motion policies and obstacle representations. 2. Integrate real-sensor data into RMP pipeline 3. Evaluate performance in simulation and on real robots.
- Highly motivated and independently working student
- Excellent programming skills in C++. Python skills would be beneficial
- Experience in robotic motion planning and/or perception would be beneficial
- Highly motivated and independently working student - Excellent programming skills in C++. Python skills would be beneficial - Experience in robotic motion planning and/or perception would be beneficial
- Michel Breyer (michel.breyer@mavt.ethz.ch)
- Nikhilesh Alatur (nikhilesh.alatur@mavt.ethz.ch)
- Michel Breyer (michel.breyer@mavt.ethz.ch) - Nikhilesh Alatur (nikhilesh.alatur@mavt.ethz.ch)