Register now After registration you will be able to apply for this opportunity online.
Safe guaranteed domain exploration with autonomous robots
In many autonomous navigation applications, the robot must interact with the environment to learn and complete tasks. Furthermore, these applications are safety-critical, and crashes cannot be afforded. This necessitates the safe learning of the unknown environment in order to achieve the task objective (e.g., detecting a leak or mapping an area). For example, consider an application of safe exploration in a warehouse with a wheeled robot to identify the source of a gas leak.
Keywords: Gaussian Processes, Active learning, Bayesian optimization, Model predictive control (MPC), Issac sim simulator
Please see the attached PDF for the complete description along with work packages.
Please see the attached PDF for the complete description along with work packages.
The problem has numerous challenges which stems from providing guarantees with unknown objectives, unknown constraints, and typically nonlinear dynamics of the robots. We have developed a theoretical framework addressing safe exploration [1]. The main objective of the project is to deploy it in more realistic simulators and extend the framework towards tasks with unknown rewards, e.g., active learning or Bayesian optimization. The developed framework will have a lot of diverse applications, such as mapping in an unknown environment, executing tasks in the presence of unknown obstacles, industrial inspection, surveillance, and more.
The problem has numerous challenges which stems from providing guarantees with unknown objectives, unknown constraints, and typically nonlinear dynamics of the robots. We have developed a theoretical framework addressing safe exploration [1]. The main objective of the project is to deploy it in more realistic simulators and extend the framework towards tasks with unknown rewards, e.g., active learning or Bayesian optimization. The developed framework will have a lot of diverse applications, such as mapping in an unknown environment, executing tasks in the presence of unknown obstacles, industrial inspection, surveillance, and more.
Manish Prajapat (manishp@ai.ethz.ch) or Johannes
Kohler (jkoehle@ethz.ch)
Manish Prajapat (manishp@ai.ethz.ch) or Johannes Kohler (jkoehle@ethz.ch)