The travelling salesman problem (TSP) is a famous NP-complete problem that asks for the shortest possible route to visit all cities on the given map. Our aim is to focus on the possible generalizations of this problem. One approach to generalize the TSP problem is to assume there is a reward associated with each city. Hence, the aim would be to maximize the collected awards within a given travelling time. Another generalization would be a case in which we have multiple salesmen. In this case, the objective would be how to distribute the tasks optimally among them such that the overall mission is satisfied or collected rewards optimal. In the end, we would like to extend the application of this problem to the case of task scheduling for a swarm of robots in which we may consider distributed optimization problems and distributed coordination as an alternative to the centralized approach.
The overall tasks in this project can be summarized as follows:
Programming skills in Python, willingness to learn ROS concepts quickly, familiarity with optimization algorithms
Our current work is building a software stack for robotics that enables programming and controlling swarms of robots on the high level, with execution verified for safety restrictions, and the project defined here would be helpful to reach some of our goals related to this robotics project.
ROS, multi-agent TSP, turtlebot, python
Rupak Majumdar, Ivan Gavran
Can environmental interference break your brakes? Find out in this project, where you will get a chance to build a distributed real-time system resembling an automotive subsystem and measure its reliability under the influence of thermal stress, radiation, and electromagnetic interference. You will first build a prototype of an automotive subsystem, e.g., a dashboard subsystem consisting of sensors, Electronic Control Units (ECUs) to input the sensor values and output the processed data, and a Controller Area Network (CAN) to network all the ECUs. You will then perform empirical measurements to evaluate how the prototype responds to harsh environmental conditions, e.g., by inducing thermal stress or by placing the system next to a microwave oven. By the end of this project, you will learn how CAN works and how real-time messages are scheduled in automotive systems, and gain experience in building distributed systems with real-time properties. All the hardware components required for the project will be provided.
Experience with team programming projects. Good C/C++ skills and willingness to work with a hardware testbed. Knowledge of real-time systems or distributed systems will be a bonus, but is not required.
Arpan Gujarati, Björn B. Brandenburg
/ Data Protection