Autonomous Systems

Autonomous systems play an increasingly important role in the field of robotics. Their technical potential is expanding rapidly, as ever powerful microprocessors, sensor systems, communication networks and energy storage is becoming available. They also have an enormous practical potential. They take over tedious or hazardous tasks, where they work better and more reliably than humans. 

One example is autonomous vehicles. They can improve traffic safety, offer individual transportation for blind or elderly people, or they simply free the driver's time. Another example is autonomous medical robots, which can help with monitoring and treating patients.

At IME, our passion is the design of intelligent algorithms for the control of autonomous systems - such as self-learning, predictive, or optimal control. We develop algorithms at the cutting-edge of modern research.

These algorithms are used to help the system with perceiving or understanding its environment, e.g., by learning from sensor data or parameter estimation. Other algorithms help the system to make intelligent decisions, perform motion planning in space and time, or to control its movements. Another important focus point of our work is on providing safety guarantees - an aspect that is particularly relevant for the two main applications we consider, autonomous vehicles and medical robots.

Autonomous Cars

Racing

As part of the lab course Robotics and Automation a car race of autonomous vehicles is held annually. The students compete each other with their self-assembled cars and their own optimized control scheme. 

Model Predictive Control

Model Predictive Control (MPC) predicts and optimizes time-varying processes over a future time horizon. MPC is used for trajectory following of a model car. It is extended to consider lane borders and obstacles on a racing track during the driving process. A toolkit for optimal control is used inside a Robotic Operating System (ROS) on the car's hardware to realize the wanted behavior in real-time.

Autonomous Drones

Drones are a challenging autonomous vehicles due to their additional degrees of freedom. On the other hand, the possible use cases are endless.

Drone Swarms

An open-source software is used to track a group of Crazyflies (mini developer drones) with a Vicon motion-capture system. The camera data is processed in a base-station PC, from where a pose array is broadcasted to all Crazyflies via 2.4 GHz radio. On the drones, external position estimates are fused with the state estimate from the onboard Inertia Measurement Unit (IMU). Additional to the position estimates, commands or trajectories can be broadcasted to the Crazyflies.
The motion-capture system provides high accuracy position data with mean errors below 2cm and low latency, allowing formations with 30cm or less distance between the drones. While four markers per drone are usually necessary, recent insights allow the use of a single marker, relying completely on the onboard IMU for orientation estimation.

Video footage of the drones in action can be found here.

Drones in Emergency Care

In the publicly funded research project "MOMENTUM - Mobile Medical technologies for Integrated Emergency Care and Medicine" suitable concepts and technologies for networking emergency medical treatment along the process chain from the point of care to the hospital are developed and evaluated. In the emergency care scenario drones are used to automatically measure the network performance in order to relay the signal and to increase the range of a mobile cell. In this way it is possible to overcome radio gaps due to coverage in rural areas or due to blockage in urban areas, e.g. by buildings. 

Further information about the Project (in German) can be found here: https://www.forschung-it-sicherheit-kommunikationssysteme.de/projekte/momentum

Projects and Theses

Click here to check out our current research projects and theses topics.