Traditional industrial assembly lines involve dozens, sometimes hundreds,of robotic manipulators. However, often, each manipulator is devoted to one specific step of the assembly process, e.g., to insert one piece or screw two pieces together. The subsequent step can only start after all previous steps are finished. Thus, the slowest step and the step with the highest failure rate directly represent the bottleneck of the assembly line.
Conversely, to the above scenario, we are interested in a more dynamic setup. In particular: The manipulators are generic and flexible. They can potentially learn to perform numerous manipulation skills; the manipulators are located within a close vicinity such that one manipulator can take-over the job of another if needed; the assembly tasks can change during run time according to e.g., online orders. With such a setup, we foresee a dynamic assembly line that is more robust to component failures and more efficient in task accomplishment.
This project will address one of the core problems: Hierarchical Task and Motion Planning (TAMP) for Manipulation Robotics. The area of TAMP attempts to improve the synergies between high-level task planning and low-level motion planning. Different approaches have been proposed toward this direction among which, e.g., hierarchical reinforcement learning and model-based task planning.
* Conscientious coordination: During your assignment, you read related literature of Learning from Demonstration (LfD) and you test the existing software for LfD.
* Observe, and think ahead: You research and develop theories on combing LfD with Task planning.
* Think holistically: The implementation of the developed theories and testing the theories and software on lab robots falls within your area of responsibility.
* Integrated implementation: You document the methods and results as scientific publications.
* Education: Master studies in the field of computer science, engineering, math or physics with very good grades, especially in Math and Engineering
* Character and Working Practice: Motivated, organized and careful
* Experience and Knowledge: Strong knowledge in Control Theory and good knowledge about Robotics; experiences in programming (Python, C++, ...)
* Language: Fluent in English
Start: November 2019
Duration: 6 months
Requirement for this internship is the enrollment at university. Please attach a motivation letter, your CV, transcript of records, enrollment certificate, examination regulations and if indicated a valid work and residence permit.
Need further information about the job?
Meng Guo (Business Department)
+4x xxx xxx xxxx5
Videos To Watch